1 Descriptive Statistics

Table 1 shows the number of Labour MPs elected in each general election from 1997 to 2015, including newly elected MPs (the ``intake’’), the number of newly elected MPs from all women shortlists (AWS), and the number of candidates selected through all women shortlists. Data in Table 1 is from the House of Commons Library (Audickas, Hawkins, & Cracknell, 2017; Kelly & White, 2016). All women shortlists were not used by Labour during the 2001 General Election.

Labour MPs and Intakes
General Election Total MPs Labour MPs Female Labour MPs Labour MPs Intake Intake Women Intake Shortlist Nominated Shortlist
1997 659 418 101 (24%) 177 64 (36%) 35 38
2001 659 412 95 (23%) 38 4 (11%) 0 0
2005 646 355 98 (28%) 40 26 (65%) 23 30
2010 650 258 81 (31%) 64 32 (50%) 28 63
2015 650 232 99 (43%) 49 31 (63%) 31 77

Table 2 shows the total size of the dataset in speeches and words by each party, including by gender for each party, and in the case of female Labour MPs, by AWS status. Details on inclusion criteria are given below.

Number of Speeches and Words in Dataset
Gender Speeches Words
All 657,547 239,123,685
Female 149,805 56,589,501
Male 507,742 182,534,184
Conservatives
All 285,308 96,186,824
Female 48,771 15,779,116
Male 236,537 80,407,708
Labour
All 262,000 99,986,437
Female 84,615 34,159,304
Non-All Women Shortlists 28,653 11,623,184
All Women Shortlists 55,962 22,536,120
Male 177,385 65,827,133
Liberal Democrat
All 72,719 28,947,968
Female 7,552 3,232,822
Male 65,167 25,715,146
Other
All 37,520 14,002,456
Female 8,867 3,418,259
Male 28,653 10,584,197

2 Methodology

Previous research on gender differences in political speech patterns has focused on differences between male and female politicians (Yu, 2014) or on variations in Hilary Clinton’s speech patterns (Bligh, Merolla, Schroedel, & Gonzalez, 2010; Jones, 2016). This paper focuses on differences in speech patterns between female Labour MPs nominated through All Women Shortlists (AWS) and female Labour MPs nominated through open shortlists. We examined differences in speaking styles using the Linguistic Inquiry and Word Count 2015 (LIWC) dictionary (Pennebaker, Boyd, Jordan, & Blackburn, 2015) and the spaCy (Honnibal & Montani, 2017) Parts-of-Speech (POS) tagger. We examined differences in the topics discussed by AWS and non-AWS MPs, using \({\chi}^2\) tests for individual words and for bigrams. We trained a Naive Bayes classifier to distinguish AWS and non-AWS speeches. We used structured topic models (STM) to identify the topics discussed by AWS and non-AWS MPs.

To account for the possible effects of age, parliamentary experience and cohort, and in order to compare women selected through all women shortlists to women who were not (but who theoretically had the opportunity to contest all-women shortlists), our analysis is been restricted only to Labour MPs first elected to the House of Commons in the 1997 General Election, up to but excluding the 2017 General Election. Comparisons between MPs of different parties are also restricted to MPs first elected in the 1997 General Election, and before the 2017 General Election. Speeches made by the Speaker, including Deputy Speakers, were also excluded. Words contained in parentheses were removed, as they are added by Hansard to provide additional information not actually spoken by the MP.1 Speeches and data on MPs’ gender and party affiliation are from a previously assembled dataset (Odell, 2018). Information on candidates selected through all women shortlists is from the House of Commons Library (Kelly & White, 2016). Unsuccessful General Election candidates selected through all women shortlists who were subsequently elected in a byelection are classified as having been selected on an all women shortlist, regardless of the selection process for that byelection. Speeches made by MPs while suspended from the Labour party where classified the same as if they had not been suspended. The dataset includes 408 different Labour MPs, 167 female MPs, 119 elected from All Women Shortlists and 48 elected from open shortlists, along with 241 male MPs.

3 Results

3.1 Linguistic Inquiry and Word Count

Word classification used the Linguistic Inquiry and Word Count 2015 (LIWC) dictionary (Pennebaker et al., 2015) and tokenising tools from the Quanteda R package (Benoit, 2018). Word counts and words-per-sentence, and calculations for determining grade level (Kincaid, Fishburne, Rogers, & Chissom, 1975) were produced using stringi (Gagolewski, 2018), an R wrapper to the ICU regex library.

Following research by Yu (2014) and Newman, Groom, Handelman, & Pennebaker (2008) on gender differences in language, we focused on the following LIWC categories to compare MPs’ speeches:

  • All Pronouns (pronoun)
  • First person singular pronouns (i)
  • First person plural pronouns (we)
  • Verbs (verb)
  • Auxiliary verbs (auxverb)
  • Social processes (social)
  • Positive emotions (posemo)
  • Negative emotions (negemo)
  • Tentative words (tentat)
  • Articles (article)
  • Prepositions (preps)
  • Anger words (anger)
  • Swear words (swear)
  • Cognitive processes (cogproc)
  • Words longer than six letters (Sixltr)

We also included mean words-per-sentence (WPS), total speach word count (WC) and Flesch–Kincaid grade level (FK) (Kincaid et al., 1975), calculated using the Quanteda (Benoit, 2018) and stringi (Gagolewski, 2018) R packages.

3.1.1 Women vs Men

Effect Sizes for Male and Female Labour MPs
Women
Men
Effect Size
Mean SD Mean SD Cohen’s d Magnitude
All Pronouns 10.07 4.60 10.15 4.99 0.02 negligible
First person singular pronouns 1.89 2.41 2.02 2.55 0.05 negligible
First person plural pronouns 0.98 1.42 0.99 1.51 0.01 negligible
Verbs 12.82 5.00 12.67 5.35 -0.03 negligible
Auxiliary verbs 7.91 3.45 7.93 3.69 0.01 negligible
Social processes 8.46 4.81 8.16 5.11 -0.06 negligible
Positive emotions 2.73 2.49 2.57 2.54 -0.06 negligible
Negative emotions 1.15 1.69 1.07 1.78 -0.05 negligible
Tentative words 1.48 1.74 1.58 1.90 0.05 negligible
More than six letters 10.62 3.67 10.26 3.92 -0.10 negligible
Articles 7.64 3.30 7.96 3.55 0.09 negligible
Prepositions 12.58 4.41 12.14 4.73 -0.10 negligible
Anger words 0.23 0.83 0.24 0.79 0.01 negligible
Swear words 0.00 0.06 0.00 0.09 0.01 negligible
Cognitive processes 8.68 4.83 8.82 5.15 0.03 negligible
Words per Sentence 43.99 19.92 41.43 20.30 -0.13 negligible
Total Word Count 402.72 691.10 370.13 647.25 -0.05 negligible
Flesh-Kincaid Grade Level 10.97 7.77 9.91 7.96 -0.13 negligible

There are no categories where gender differences meet the effect size threshold of \(|0.2|\) suggested by Cohen (1988, pp. 25–26) to indicate a small effect. 4 categories – words with more than six letters, prepositions, words-per-sentence and Flesh-Kincaid grade level – met or exceeded the \(|0.1|\) threshold suggested by Newman et al. (2008).

3.1.2 Shortlists vs Non-Shortlists

Figure shows changes in the occurences of selected LIWC terms, words-per-sentence, total word count and Flesch–Kincaid grade level, over the course of an MP’s career, as measured since the time an MP was first elected. There do not appear to be any notable changes in speaking style over the course of female Labour MPs’ careers. Figure shows changes in the occurences of the same selected terms from 1997–2017. As in Figure , there do not appear to be any meaningful trends in the use of the selected terms over time.

\label{sl-key-y-since-start}Occurence of selected LIWC terms, by time as MP

Occurence of selected LIWC terms, by time as MP

\label{sl-key-date}Occurence of selected LIWC terms, by date

Occurence of selected LIWC terms, by date

Effect Sizes for Female Labour MPs by selection process
All Women Shortlists
Open Shortlists
Effect Size
Mean SD Mean SD Cohen’s d Magnitude
All Pronouns 10.01 4.66 10.19 4.47 -0.04 negligible
First person singular pronouns 1.86 2.41 1.95 2.42 -0.04 negligible
First person plural pronouns 0.88 1.36 1.16 1.51 -0.19 negligible
Verbs 12.88 5.10 12.69 4.80 0.04 negligible
Auxiliary verbs 7.94 3.49 7.86 3.38 0.02 negligible
Social processes 8.46 4.93 8.44 4.58 0.00 negligible
Positive emotions 2.69 2.52 2.81 2.42 -0.05 negligible
Negative emotions 1.17 1.69 1.13 1.68 0.02 negligible
Tentative words 1.48 1.75 1.49 1.73 0.00 negligible
More than six letters 10.56 3.72 10.74 3.56 -0.05 negligible
Articles 7.69 3.38 7.55 3.14 0.04 negligible
Prepositions 12.55 4.54 12.63 4.14 -0.02 negligible
Anger words 0.23 0.79 0.24 0.91 -0.01 negligible
Swear words 0.00 0.06 0.00 0.05 0.01 negligible
Cognitive processes 8.59 4.90 8.85 4.69 -0.05 negligible
Words per Sentence 44.39 20.69 43.21 18.31 0.06 negligible
Total Word Count 401.70 704.15 404.73 664.87 0.00 negligible
Flesh-Kincaid Grade Level 11.13 8.06 10.64 7.15 0.07 negligible

There are no categories among female Labour MPs by selection process meeting the \(|0.2|\) threshold. Only one category – first person plural pronouns, d=0.19 – exceeded \(|0.1|\).

3.1.3 Conservatives vs Labour

Effect Sizes for All Labour and Conservative MPs
Labour
Conservatives
Effect Size
Mean SD Mean SD Cohen’s d Magnitude
All Pronouns 10.12 4.87 10.61 4.84 0.10 negligible
First person singular pronouns 1.98 2.51 2.14 2.56 0.06 negligible
First person plural pronouns 0.98 1.48 1.22 1.70 0.15 negligible
Verbs 12.72 5.24 12.92 5.14 0.04 negligible
Auxiliary verbs 7.93 3.61 8.16 3.58 0.06 negligible
Social processes 8.26 5.02 8.11 4.80 -0.03 negligible
Positive emotions 2.63 2.52 2.85 2.66 0.09 negligible
Negative emotions 1.10 1.75 1.04 1.79 -0.03 negligible
Tentative words 1.55 1.85 1.57 1.88 0.01 negligible
More than six letters 10.38 3.84 10.31 3.75 -0.02 negligible
Articles 7.86 3.47 7.81 3.45 -0.01 negligible
Prepositions 12.28 4.63 12.35 4.49 0.02 negligible
Anger words 0.24 0.80 0.24 0.82 0.00 negligible
Swear words 0.00 0.08 0.00 0.10 0.00 negligible
Cognitive processes 8.77 5.05 8.85 5.06 0.01 negligible
Words per Sentence 42.26 20.22 43.07 20.39 0.04 negligible
Total Word Count 380.64 661.91 336.23 594.06 -0.07 negligible
Flesh-Kincaid Grade Level 10.25 7.91 10.54 7.99 0.04 negligible

There are no categories with effect sizes exceeding \(|0.2|\) between Labour and Conservative MPs, and only one (first person plural pronouns) exceeding \(|0.1|\).

3.1.4 All MPs Gender Differences

There are no categories with effect sizes exceeding \(|0.2|\) when comparing all male and female MPs elected from 1997 onwards. There is only one category, “Articles”, with an effect size of 0.11, greater than the \(|0.1|\) threshold suggested by Newman et al. (2008).

Effect Sizes for Male and Female MPs, All Parties
Women
Men
Effect Size
Mean SD Mean SD Cohen’s d Magnitude
All Pronouns 10.31 4.65 10.26 4.90 -0.01 negligible
First person singular pronouns 1.99 2.45 2.00 2.52 0.00 negligible
First person plural pronouns 1.11 1.57 1.08 1.59 -0.02 negligible
Verbs 12.89 4.98 12.80 5.26 -0.02 negligible
Auxiliary verbs 8.01 3.45 8.08 3.64 0.02 negligible
Social processes 8.44 4.77 7.99 4.92 -0.09 negligible
Positive emotions 2.84 2.53 2.70 2.58 -0.06 negligible
Negative emotions 1.10 1.65 1.07 1.78 -0.01 negligible
Tentative words 1.47 1.73 1.61 1.91 0.08 negligible
More than six letters 10.57 3.66 10.34 3.83 -0.06 negligible
Articles 7.63 3.30 8.00 3.51 0.11 negligible
Prepositions 12.59 4.36 12.22 4.61 -0.08 negligible
Anger words 0.23 0.79 0.25 0.82 0.02 negligible
Swear words 0.00 0.05 0.00 0.10 0.01 negligible
Cognitive processes 8.68 4.80 8.93 5.12 0.05 negligible
Words per Sentence 44.00 20.02 42.69 20.65 -0.07 negligible
Total Word Count 376.81 648.62 358.56 624.84 -0.03 negligible
Flesh-Kincaid Grade Level 10.95 7.82 10.43 8.08 -0.07 negligible

3.2 POS Analysis

Part-of-Speech Effect Sizes for Male and Female Labour MPs
Women
Men
Effect Size
Word Type Mean SD Mean SD Cohen’s d Magnitude
All Nouns 22.17 9.56 21.67 10.92 -0.05 negligible
Plural Nouns 5.86 3.71 5.04 3.79 -0.22 small
Singular Nouns 15.61 9.81 16.01 11.16 0.04 negligible
Adjectives 9.58 4.77 9.28 5.29 -0.06 negligible
Adverbs 4.91 4.25 5.06 4.91 0.04 negligible
Verbs 20.97 9.52 20.81 10.28 -0.02 negligible
Part-of-Speech Effect Sizes for AWS and non-AWS Labour MPs
All Women Shortlists
Open Shorlists
Effect Size
Word Type Mean SD Mean SD Cohen’s d Magnitude
All Nouns 22.16 8.72 22.18 9.97 -0.04 negligible
Plural Nouns 6.03 3.59 5.77 3.76 -0.16 negligible
Singular Nouns 15.50 8.93 15.67 10.23 0.03 negligible
Adjectives 9.83 4.58 9.45 4.86 -0.02 negligible
Adverbs 4.95 3.76 4.89 4.49 0.03 negligible
Verbs 20.92 9.02 21.00 9.77 -0.02 negligible

Part-of-speech (POS) tagging was done using the spaCy library (Honnibal & Montani, 2017) and the spacyr package (Benoit & Matsuo, 2018). There is one small gender difference (d = -0.22) in the use of plural nouns, which make up 5.86% of the words used by female Labour MPs, compared to 5.04% of words spoken by male Labour MPs. As with LIWC, there are no categories where d >= \(|0.2|\) when comparing female Labour MPs by selection process, and only one category – plural nouns – with an effect size of d >= \(|0.1|\).

3.3 Keyness

We calculated the keyness of words to identify gender differences in the choices of topics raised and terminology used by both male and female Labour MPs, and by short-list and non-shortlist female Labour MPs. We have also calculated keyness between Labour and Conservative MPs for the purposes of illustration. All keyness figures include the 25 most disproportionately common words among each group, as determined by \({\chi}^2\) tests using quanteda (Benoit, 2018).

3.3.1 Labour Men vs Women

Keyness – a linguistic measure of the frequency of different words in two groups of texts – reveals clear gender differences in the most disproportionately common words used by female and male Labour MPs, illustrated in Figure .

Unsurprisingly, despite male MPs saying almost twice as many words (65,827,133 vs 34,159,304) as their female colleagues, female Labour MPs were more than two-and-a-half (2.61) times as likely to say “women”. They were also much more likely to use “women’s” and “woman” in parliamentary debate. Female Labour MPs also appear much more likely to discuss “children”, “people”, “care”, “families”, “home”, “parents”, “work” and social policy areas such as “services”, “disabled [people]” and “housing” than their male colleagues. Male MPs were more likely to refer to military topics (“Iraq”, “nuclear”), and to parliamentary process and protocol – “question”, “political”, “conservative”, “electoral”, “house”, “party”, “argument” “liberal” and “point” are far more common in speeches by male Labour MPs than by female ones. This could suggest that male Labour MPs are more comfortable using the traditional language of House of Commons debate, and are more concerned with the rules, procedures and processes of the parliamentary system than their female colleagues.

\label{gender-keyness}Keyness between Labour MPs, by Gender

Keyness between Labour MPs, by Gender

3.3.2 Shortlists vs Non-Shortlists

Keyness differences by selection process (Figure ) are not as obviously stereotypical. Nonetheless, the most common words amongst AWS MPs included “carers”, “disabled”, “bedroom” and “sen” (Special Educational Needs). Also of note is AWS MPs making more references to their “constituency” and its “constituents”, suggesting that AWS MPs may draw more heavily on the fact they were elected by their constituents as a source political legitimacy, or are more likely to illustrate a point with an example from their constitutency, compared to non-AWS MPs.

\label{sl-keyness}Keyness between Female Labour MPs, by Selection Process

Keyness between Female Labour MPs, by Selection Process

3.3.3 Labour vs Conservative

The keyness differences (Figure ) between Labour and Conservative MPs are much greater than gender or AWS differences within Labour. The very high use of “Lady” by Conservative MPs is reflective of the greater proportion of female MPs in other parties, as it is often used to refer to comments by other members of the house. It may also represent a greater use of traditional hosue decorum by Conservative MPs.

\label{party-keyness}Keyness between Labour and Conservative MPs

Keyness between Labour and Conservative MPs

3.4 Bigrams

We created bigrams of all first person plural and singular pronouns for female Labour MPs. As above, AWS MPs are far more likely to make references to their constituency or their constituents.

\label{bigrams-keyness}Bigram Keyness in Female Labour MPs by Selection Process

Bigram Keyness in Female Labour MPs by Selection Process

3.5 Naive Bayes classification

We trained a Naive Bayes classifier with document-frequency priors and a multinomial distribution to predict the gender of speakers when given speeches by all Labour MPs in our dataset, and the selection process when only given female Labour MPs. The accuracy of both models were roughly equivalent, 70.67% accuracy when predicting gender and 71.22% when predicting shortlists. By contrast, the classifier could distinguish between Labour and Conservative speeches with 74.23% accuracy.

3.6 Topic Models

Using topic models to classify text is widely used in social sciences (Grimmer & Stewart, 2013), as, when combined with the large volume of plain text data available, it allows for a rapid and consistent method of analysis . Topic modelling and other statistic methods of textual analysis are not a substitute for reading the texts themselves, but can augment other analysis or – as in this case – analyse and classify larger amounts of text than would be feasible using human coders (Grimmer & Stewart, 2013). Topic models classify a series of documents (in this case individual speeches) into one of a given number of topics, identifying terms that are common in some documents but rare in others. When developing topic models, there is a trade-off between high precision in the classification of each document with broader topics when using smaller numbers of topics, or lower precision in individual speech classification with more finely-grained topics when using larger numbers of topics. Grimmer & Stewart (2013) also highlight the importance of validating unsurpervised topic models when applied to new sets of texts, which we have done below.

The R package stm (Roberts, Stewart, & Tingley, 2018) implements a structured topic model (STM) (Arora et al., 2013; Roberts, Stewart, & Airoldi, 2016). An STM incorporates covariates into the topic classification algorithm, creating possibilities for hypothesis testing. This differs from traditional topic modelling methods using latent variables to identify topics (e.g. with latent Dirichlet allocation Blei, Ng, & Jordan, 2003), and then comparing proportions of each topic to one or more external variables. STM allows us to incorporate the variables we are interested in to the topic model itself using a generalised linear model; i.e. the proportion of speechs classified as belonging to each topic can vary as a function of the AWS and gender variables.

We incorporated the AWS status of speakers and their gender as prevalence covariates into our topic model.

\label{topic-model-selection}Topic Model Selection

Topic Model Selection

We created six topic models with different numbers of topics (K). We created models with 30, 45, 60, 80 and 100 topics, and used an algorithm developed by Lee & Mimno (2014), implemented in the stm package (Roberts et al., 2018), which resulted in K = 66. Figure shows, clockwise from the top-left, heldout likelihood [explain], lower bound [explain], semantic coherence (Mimno, Wallach, Talley, Leenders, & McCallum, 2011), and the multinomial dispersion of the STM residuals (Taddy, 2012),

As seen in Figure , the K = 66 result appears to produce the best result, a topic model with 66 topics, across 251,072 speeches with a dictionary of 241,625 words. All models were created using the “spectral” method developed by Arora et al. (2013), implemented in the stm package by Roberts et al. (2018).

One of the topics – Topic 66 – is never the most likely topic in the matrix of number of documents by number of topics – labelled \(\theta\) by Roberts et al. (2018) – and so while it is included in the model, assignment of single topics to speeches uses the highest \(\theta\) for each speech. Other topics are rarely used – Topic 53, which we labelled “Dispatch Box”, only has five topics assigned to it, four from Male MPs and one from an AWS MP.

Figure is a Fruchterman-Reingold force-directed diagram (Fruchterman & Reingold, 1991) of correlations between different topics. Larger vertices indicate more common topics, and the colour scale indicates the proportion of speeches classed in that topic made by AWS and non-AWS female Labour MPs, respectively. Edges indicate positive correlations between the two linked topics.

\label{k0-network}Fruchterman-Reingold plot of Topic Network

Fruchterman-Reingold plot of Topic Network

The stm package includes the estimateEffect function, which creates a regression model (Table 9) using individual documents (speeches) as observations, with the proportion of a each document fitting each topic as the dependent variable and model covariates (AWS status and gender) as independent variables. The intercept in this model is all speeches by male Labour MPs.

Topic Estimates
Estimate Standard Error t value Pr(>|t|)
Topic 1 – Employment & unions
Intercept 0.0120878 0.0001165 103.7197503 < 0.001 ***
Non-AWS -0.0003810 0.0003152 -1.2086545 0.23
AWS -0.0013425 0.0002462 -5.4524847 < 0.001 ***
Topic 2 – Legal system
Intercept 0.0167073 0.0001807 92.4456042 < 0.001 ***
Non-AWS 0.0069619 0.0005414 12.8594336 < 0.001 ***
AWS -0.0033127 0.0003284 -10.0881608 < 0.001 ***
Topic 3 – Roads
Intercept 0.0116636 0.0001539 75.8065626 < 0.001 ***
Non-AWS -0.0014880 0.0004093 -3.6354260 < 0.001 ***
AWS -0.0019521 0.0002977 -6.5568837 < 0.001 ***
Topic 4 – Housing
Intercept 0.0112824 0.0001697 66.5002232 < 0.001 ***
Non-AWS 0.0044584 0.0004867 9.1606890 < 0.001 ***
AWS 0.0060412 0.0003717 16.2509038 < 0.001 ***
Topic 5 – Police, firefighters & prison
Intercept 0.0140708 0.0001772 79.4090929 < 0.001 ***
Non-AWS 0.0032592 0.0005277 6.1760547 < 0.001 ***
AWS -0.0003215 0.0003600 -0.8929765 0.37
Topic 6 – Northern Ireland
Intercept 0.0089511 0.0000477 187.6762538 < 0.001 ***
Non-AWS 0.0000909 0.0001253 0.7256671 0.47
AWS -0.0003738 0.0001131 -3.3055045 < 0.001 ***
Topic 7 – Committee
Intercept 0.0213275 0.0001402 152.0808233 < 0.001 ***
Non-AWS -0.0007053 0.0003778 -1.8665872 0.062
AWS -0.0019513 0.0002695 -7.2414395 < 0.001 ***
Topic 8 – Schools
Intercept 0.0147216 0.0001998 73.6946174 < 0.001 ***
Non-AWS -0.0009629 0.0005030 -1.9144999 0.056
AWS 0.0021239 0.0004180 5.0810609 < 0.001 ***
Topic 9 – Energy & climate change
Intercept 0.0170599 0.0001997 85.4395693 < 0.001 ***
Non-AWS -0.0011701 0.0005240 -2.2331311 0.026
AWS -0.0035162 0.0004352 -8.0791703 < 0.001 ***
Topic 10 – Defence
Intercept 0.0157879 0.0001960 80.5622700 < 0.001 ***
Non-AWS -0.0075466 0.0004666 -16.1743503 < 0.001 ***
AWS -0.0054234 0.0003707 -14.6290075 < 0.001 ***
Topic 11 – Parliament
Intercept 0.0118979 0.0000780 152.4772016 < 0.001 ***
Non-AWS -0.0036949 0.0002003 -18.4467310 < 0.001 ***
AWS -0.0010964 0.0001547 -7.0857345 < 0.001 ***
Topic 12 – International politics
Intercept 0.0126075 0.0001309 96.3168538 < 0.001 ***
Non-AWS -0.0042329 0.0003197 -13.2416445 < 0.001 ***
AWS -0.0054763 0.0002592 -21.1237522 < 0.001 ***
Topic 13 – Ministers
Intercept 0.0167421 0.0001102 151.8882491 < 0.001 ***
Non-AWS -0.0029730 0.0002828 -10.5109108 < 0.001 ***
AWS 0.0031466 0.0002352 13.3774397 < 0.001 ***
Topic 14 – Policy impact
Intercept 0.0115302 0.0000459 251.4641894 < 0.001 ***
Non-AWS 0.0002485 0.0001403 1.7710959 0.077
AWS 0.0013681 0.0001036 13.2044108 < 0.001 ***
Topic 15 – Gender
Intercept 0.0048727 0.0001177 41.4094648 < 0.001 ***
Non-AWS 0.0123737 0.0003762 32.8913879 < 0.001 ***
AWS 0.0119861 0.0003404 35.2071183 < 0.001 ***
Topic 16 – Regional development
Intercept 0.0230389 0.0001294 178.0930817 < 0.001 ***
Non-AWS 0.0070464 0.0003591 19.6216388 < 0.001 ***
AWS 0.0002652 0.0002577 1.0290190 0.30
Topic 17 – Communications
Intercept 0.0097576 0.0001219 80.0144325 < 0.001 ***
Non-AWS -0.0006793 0.0003576 -1.8994880 0.058
AWS -0.0011976 0.0002649 -4.5216601 < 0.001 ***
Topic 18 – Immigration
Intercept 0.0087069 0.0000954 91.2603658 < 0.001 ***
Non-AWS 0.0007386 0.0002701 2.7352156 0.006 **
AWS -0.0004153 0.0001896 -2.1908890 0.028
Topic 19 – Health system
Intercept 0.0161551 0.0002162 74.7090681 < 0.001 ***
Non-AWS 0.0112524 0.0006432 17.4956211 < 0.001 ***
AWS 0.0062993 0.0004773 13.1973842 < 0.001 ***
Topic 20 – International development
Intercept 0.0160713 0.0001993 80.6278360 < 0.001 ***
Non-AWS 0.0004194 0.0005254 0.7981813 0.42
AWS -0.0033527 0.0003846 -8.7167408 < 0.001 ***
Topic 21 – Benefits & disability
Intercept 0.0120320 0.0001424 84.4678321 < 0.001 ***
Non-AWS 0.0009218 0.0003803 2.4241165 0.015
AWS 0.0120279 0.0003149 38.1929006 < 0.001 ***
Topic 22 – Sport & culture
Intercept 0.0127190 0.0001621 78.4688321 < 0.001 ***
Non-AWS -0.0024706 0.0004138 -5.9703426 < 0.001 ***
AWS 0.0007447 0.0003267 2.2795017 0.023
Topic 23 – History
Intercept 0.0137433 0.0001078 127.4899255 < 0.001 ***
Non-AWS -0.0060891 0.0002707 -22.4939047 < 0.001 ***
AWS -0.0040079 0.0002050 -19.5533523 < 0.001 ***
Topic 24 – Higher education & skills
Intercept 0.0143143 0.0001657 86.3679689 < 0.001 ***
Non-AWS -0.0010161 0.0004360 -2.3308295 0.020
AWS -0.0001187 0.0003382 -0.3510073 0.73
Topic 25 – Concurring point
Intercept 0.0155251 0.0000456 340.1375115 < 0.001 ***
Non-AWS -0.0018971 0.0001193 -15.8982760 < 0.001 ***
AWS -0.0030027 0.0000874 -34.3703617 < 0.001 ***
Topic 26 – Pensions
Intercept 0.0146930 0.0001644 89.3671042 < 0.001 ***
Non-AWS 0.0007041 0.0004244 1.6590969 0.097
AWS 0.0026241 0.0003283 7.9921504 < 0.001 ***
Topic 27 – Points of order
Intercept 0.0177836 0.0001285 138.4473569 < 0.001 ***
Non-AWS -0.0065323 0.0003178 -20.5548984 < 0.001 ***
AWS -0.0048141 0.0002487 -19.3602209 < 0.001 ***
Topic 28 – Issues
Intercept 0.0344877 0.0000996 346.3714353 < 0.001 ***
Non-AWS 0.0070247 0.0002806 25.0317360 < 0.001 ***
AWS -0.0025873 0.0001991 -12.9931954 < 0.001 ***
Topic 29 – Constituencies
Intercept 0.0131812 0.0000489 269.3165720 < 0.001 ***
Non-AWS 0.0011058 0.0001418 7.7965760 < 0.001 ***
AWS 0.0029674 0.0001067 27.8022737 < 0.001 ***
Topic 30 – Ethnic groups & racism
Intercept 0.0085786 0.0000747 114.8716095 < 0.001 ***
Non-AWS 0.0019097 0.0002190 8.7201118 < 0.001 ***
AWS 0.0019263 0.0001706 11.2936595 < 0.001 ***
Topic 31 – Amendments
Intercept 0.0149892 0.0001578 95.0116017 < 0.001 ***
Non-AWS -0.0017682 0.0004272 -4.1393215 < 0.001 ***
AWS -0.0033163 0.0003278 -10.1159883 < 0.001 ***
Topic 32 – Reports
Intercept 0.0169549 0.0001055 160.7352492 < 0.001 ***
Non-AWS 0.0012172 0.0002906 4.1890044 < 0.001 ***
AWS 0.0013432 0.0002378 5.6495257 < 0.001 ***
Topic 33 – People
Intercept 0.0377541 0.0001140 331.1922896 < 0.001 ***
Non-AWS -0.0022830 0.0002853 -8.0020366 < 0.001 ***
AWS -0.0010471 0.0002453 -4.2683701 < 0.001 ***
Topic 34 – Wales & Scotland
Intercept 0.0135424 0.0001625 83.3351438 < 0.001 ***
Non-AWS -0.0047705 0.0003679 -12.9658978 < 0.001 ***
AWS -0.0023222 0.0003020 -7.6899974 < 0.001 ***
Topic 35 – Alcohol & tobacco
Intercept 0.0108961 0.0001606 67.8366113 < 0.001 ***
Non-AWS -0.0008367 0.0004297 -1.9470649 0.052
AWS 0.0011919 0.0003144 3.7912405 < 0.001 ***
Topic 36 – Place names
Intercept 0.0083691 0.0000668 125.3348056 < 0.001 ***
Non-AWS 0.0000201 0.0001839 0.1095173 0.91
AWS 0.0011695 0.0001439 8.1288028 < 0.001 ***
Topic 37 – Budget
Intercept 0.0246567 0.0001726 142.8440333 < 0.001 ***
Non-AWS -0.0023151 0.0004562 -5.0752828 < 0.001 ***
AWS 0.0007173 0.0003664 1.9574684 0.050
Topic 38 – Tax
Intercept 0.0193464 0.0001866 103.7045489 < 0.001 ***
Non-AWS -0.0013559 0.0005290 -2.5630049 0.010
AWS 0.0054447 0.0003805 14.3106178 < 0.001 ***
Topic 39 – Private companies
Intercept 0.0123793 0.0001253 98.7991149 < 0.001 ***
Non-AWS 0.0005567 0.0003571 1.5587397 0.12
AWS -0.0017965 0.0002437 -7.3725615 < 0.001 ***
Topic 40 – Environment & fishing
Intercept 0.0094611 0.0001529 61.8773179 < 0.001 ***
Non-AWS -0.0030998 0.0003545 -8.7447616 < 0.001 ***
AWS -0.0021435 0.0002943 -7.2823198 < 0.001 ***
Topic 41 – Crime
Intercept 0.0141417 0.0001685 83.9390282 < 0.001 ***
Non-AWS 0.0086053 0.0005396 15.9482423 < 0.001 ***
AWS 0.0034705 0.0003617 9.5959598 < 0.001 ***
Topic 42 – Bills
Intercept 0.0244469 0.0001477 165.5088933 < 0.001 ***
Non-AWS 0.0021402 0.0004131 5.1813793 < 0.001 ***
AWS -0.0029690 0.0002777 -10.6907378 < 0.001 ***
Topic 43 – Children
Intercept 0.0076748 0.0001318 58.2192133 < 0.001 ***
Non-AWS 0.0092060 0.0004020 22.9002073 < 0.001 ***
AWS 0.0095695 0.0002785 34.3604552 < 0.001 ***
Topic 44 – Utilities & PFI
Intercept 0.0123343 0.0000947 130.2237407 < 0.001 ***
Non-AWS -0.0007793 0.0002336 -3.3361768 < 0.001 ***
AWS 0.0002432 0.0001882 1.2923235 0.20
Topic 45 – Middle East
Intercept 0.0174941 0.0002070 84.5105955 < 0.001 ***
Non-AWS -0.0028422 0.0005201 -5.4643415 < 0.001 ***
AWS -0.0017110 0.0004323 -3.9577373 < 0.001 ***
Topic 46 – Local authorities
Intercept 0.0179670 0.0001428 125.8631815 < 0.001 ***
Non-AWS 0.0044516 0.0004091 10.8810319 < 0.001 ***
AWS 0.0001235 0.0003122 0.3956712 0.69
Topic 47 – Elections
Intercept 0.0181758 0.0001773 102.5372686 < 0.001 ***
Non-AWS -0.0091680 0.0004181 -21.9262581 < 0.001 ***
AWS -0.0068076 0.0003399 -20.0272833 < 0.001 ***
Topic 48 – Debate
Intercept 0.0180048 0.0000743 242.3086768 < 0.001 ***
Non-AWS -0.0034968 0.0001991 -17.5609063 < 0.001 ***
AWS -0.0009797 0.0001463 -6.6961215 < 0.001 ***
Topic 49 – Transport
Intercept 0.0164463 0.0002009 81.8737474 < 0.001 ***
Non-AWS -0.0027555 0.0005213 -5.2862444 < 0.001 ***
AWS 0.0008788 0.0003939 2.2309711 0.026
Topic 50 – Questions
Intercept 0.0161736 0.0000751 215.4118514 < 0.001 ***
Non-AWS 0.0001342 0.0001919 0.6992798 0.48
AWS 0.0002167 0.0001602 1.3529327 0.18
Topic 51 – Families
Intercept 0.0101065 0.0001119 90.2937989 < 0.001 ***
Non-AWS 0.0019089 0.0003342 5.7125626 < 0.001 ***
AWS 0.0058701 0.0002475 23.7192595 < 0.001 ***
Topic 52 – Health research
Intercept 0.0088034 0.0001494 58.9166750 < 0.001 ***
Non-AWS 0.0076348 0.0004357 17.5222299 < 0.001 ***
AWS 0.0036106 0.0003257 11.0867708 < 0.001 ***
Topic 53 – Dispatch box
Intercept 0.0075504 0.0000224 337.1444588 < 0.001 ***
Non-AWS -0.0011342 0.0000544 -20.8552563 < 0.001 ***
AWS -0.0009564 0.0000451 -21.1903232 < 0.001 ***
Topic 54 – Parties
Intercept 0.0248210 0.0001234 201.1304045 < 0.001 ***
Non-AWS -0.0066287 0.0003392 -19.5393211 < 0.001 ***
AWS -0.0059999 0.0002627 -22.8422191 < 0.001 ***
Topic 55 – Statements
Intercept 0.0211143 0.0000690 306.1720610 < 0.001 ***
Non-AWS -0.0045075 0.0001842 -24.4766058 < 0.001 ***
AWS -0.0014974 0.0001317 -11.3713303 < 0.001 ***
Topic 56 – European Union
Intercept 0.0163470 0.0001616 101.1587437 < 0.001 ***
Non-AWS -0.0024144 0.0004584 -5.2664499 < 0.001 ***
AWS -0.0053826 0.0003343 -16.1017544 < 0.001 ***
Topic 57 – Locations
Intercept 0.0100661 0.0001091 92.2792766 < 0.001 ***
Non-AWS -0.0025097 0.0002668 -9.4073963 < 0.001 ***
AWS 0.0000332 0.0002083 0.1595476 0.87
Topic 58 – Jobs & manufacturing
Intercept 0.0175831 0.0001665 105.6181577 < 0.001 ***
Non-AWS -0.0016175 0.0004333 -3.7329769 < 0.001 ***
AWS 0.0012071 0.0003447 3.5020846 < 0.001 ***
Topic 59 – Small business
Intercept 0.0070676 0.0000732 96.5570190 < 0.001 ***
Non-AWS 0.0005514 0.0001972 2.7956892 0.005 **
AWS -0.0003676 0.0001461 -2.5164206 0.012
Topic 60 – Agreement & disagreement
Intercept 0.0328529 0.0001150 285.6388062 < 0.001 ***
Non-AWS -0.0089905 0.0003040 -29.5722122 < 0.001 ***
AWS -0.0109407 0.0002057 -53.1886055 < 0.001 ***
Topic 61 – Voluntary sector
Intercept 0.0187132 0.0001260 148.5162257 < 0.001 ***
Non-AWS 0.0111211 0.0003694 30.1062250 < 0.001 ***
AWS 0.0056575 0.0002515 22.4926245 < 0.001 ***
Topic 62 – Comments
Intercept 0.0152718 0.0000665 229.6327169 < 0.001 ***
Non-AWS -0.0029210 0.0001694 -17.2420586 < 0.001 ***
AWS -0.0040215 0.0001215 -33.0857000 < 0.001 ***
Topic 63 – Social care
Intercept 0.0090458 0.0001161 77.9441211 < 0.001 ***
Non-AWS 0.0094907 0.0003834 24.7509317 < 0.001 ***
AWS 0.0073850 0.0002788 26.4862992 < 0.001 ***
Topic 64 – Time
Intercept 0.0213817 0.0000675 316.5606944 < 0.001 ***
Non-AWS -0.0020756 0.0001747 -11.8788730 < 0.001 ***
AWS -0.0016513 0.0001438 -11.4861530 < 0.001 ***
Topic 65 – Media & animals
Intercept 0.0121376 0.0001647 73.6805837 < 0.001 ***
Non-AWS -0.0057140 0.0004053 -14.0975598 < 0.001 ***
AWS -0.0017731 0.0003155 -5.6195042 < 0.001 ***
Topic 66 – Other
Intercept 0.0038249 0.0000116 330.2711899 < 0.001 ***
Non-AWS 0.0002528 0.0000294 8.5949370 < 0.001 ***
AWS 0.0003065 0.0000250 12.2518082 < 0.001 ***

Table 10 shows the number and percentage of speeches assigned to each topic, based on its \(\theta\) value. The results in this table differ slightly from those in Table 9, as it uses a “winner-take-all” method to assign an overall topic to each speech, rather than a prevalence of a given topic across all speeches.

Count and Distribution of Topics
Topic AWS Speeches Percent of AWS Speeches Non-AWS Speeches Percent of non-AWS Speeches Male MP Speeches Percent of Male MP Speeches
  1. Employment & unions
452 0.84% 260 0.93% 2,149 1.27%
  1. Legal system
865 1.61% 1,096 3.93% 3,884 2.29%
  1. Roads
558 1.04% 298 1.07% 2,142 1.26%
  1. Housing
1,383 2.57% 665 2.39% 2,416 1.43%
  1. Police, firefighters & prison
1,046 1.94% 709 2.54% 3,353 1.98%
  1. Northern Ireland
221 0.41% 66 0.24% 603 0.36%
  1. Committee
1,050 1.95% 492 1.77% 3,888 2.29%
  1. Schools
1,367 2.54% 522 1.87% 3,780 2.23%
  1. Energy & climate change
1,105 2.05% 745 2.67% 4,630 2.73%
  1. Defence
794 1.48% 280 1.00% 3,999 2.36%
  1. Parliament
375 0.70% 85 0.31% 1,079 0.64%
  1. International politics
289 0.54% 161 0.58% 2,021 1.19%
  1. Ministers
872 1.62% 242 0.87% 2,083 1.23%
  1. Policy impact
242 0.45% 68 0.24% 417 0.25%
  1. Gender
1,257 2.34% 701 2.52% 551 0.33%
  1. Regional development
931 1.73% 710 2.55% 2,704 1.60%
  1. Communications
385 0.72% 287 1.03% 1,751 1.03%
  1. Immigration
425 0.79% 220 0.79% 1,218 0.72%
  1. Health system
2,149 4.00% 1,489 5.34% 4,682 2.76%
  1. International development
862 1.60% 687 2.47% 3,718 2.19%
  1. Benefits & disability
1,888 3.51% 317 1.14% 2,101 1.24%
  1. Sport & culture
846 1.57% 317 1.14% 2,628 1.55%
  1. History
299 0.56% 140 0.50% 1,720 1.02%
  1. Higher education & skills
974 1.81% 456 1.64% 3,501 2.07%
  1. Concurring point
33 0.06% 9 0.03% 139 0.08%
  1. Pensions
1,231 2.29% 529 1.90% 2,982 1.76%
  1. Points of order
787 1.46% 230 0.83% 4,069 2.40%
  1. Issues
1,618 3.01% 1,720 6.17% 6,745 3.98%
  1. Constituencies
125 0.23% 30 0.11% 228 0.13%
  1. Ethnic groups & racism
454 0.84% 203 0.73% 945 0.56%
  1. Amendments
526 0.98% 317 1.14% 2,293 1.35%
  1. Reports
536 1.00% 322 1.16% 1,488 0.88%
  1. People
2,818 5.24% 1,048 3.76% 9,136 5.39%
  1. Wales & Scotland
662 1.23% 224 0.80% 2,655 1.57%
  1. Alcohol & tobacco
846 1.57% 336 1.21% 2,357 1.39%
  1. Place names
163 0.30% 47 0.17% 447 0.26%
  1. Budget
1,616 3.00% 668 2.40% 5,567 3.29%
  1. Tax
2,149 4.00% 691 2.48% 4,562 2.69%
  1. Private companies
452 0.84% 362 1.30% 1,794 1.06%
  1. Environment & fishing
435 0.81% 186 0.67% 1,689 1.00%
  1. Crime
1,408 2.62% 926 3.32% 3,073 1.81%
  1. Bills
1,199 2.23% 931 3.34% 4,534 2.68%
  1. Children
1,176 2.19% 631 2.26% 1,298 0.77%
  1. Utilities & PFI
433 0.81% 175 0.63% 1,416 0.84%
  1. Middle East
1,284 2.39% 588 2.11% 4,543 2.68%
  1. Local authorities
1,050 1.95% 711 2.55% 3,686 2.18%
  1. Elections
759 1.41% 240 0.86% 4,308 2.54%
  1. Debate
422 0.78% 128 0.46% 1,364 0.81%
  1. Transport
1,517 2.82% 546 1.96% 4,172 2.46%
  1. Questions
390 0.73% 182 0.65% 1,115 0.66%
  1. Families
786 1.46% 276 0.99% 1,169 0.69%
  1. Health research
743 1.38% 591 2.12% 1,467 0.87%
  1. Dispatch box
1 0.00% NA NA% 4 0.00%
  1. Parties
879 1.63% 438 1.57% 5,053 2.98%
  1. Statements
180 0.33% 79 0.28% 856 0.51%
  1. European Union
769 1.43% 554 1.99% 3,949 2.33%
  1. Locations
299 0.56% 126 0.45% 1,112 0.66%
  1. Jobs & manufacturing
1,426 2.65% 586 2.10% 4,162 2.46%
  1. Small business
229 0.43% 183 0.66% 791 0.47%
  1. Agreement & disagreement
523 0.97% 275 0.99% 4,962 2.93%
  1. Voluntary sector
1,307 2.43% 853 3.06% 2,480 1.46%
  1. Comments
108 0.20% 95 0.34% 865 0.51%
  1. Social care
865 1.61% 521 1.87% 1,187 0.70%
  1. Time
208 0.39% 103 0.37% 930 0.55%
  1. Media & animals
741 1.38% 190 0.68% 2,811 1.66%

3.6.1 Topic Graphs

The estimate effects in these graphs were extracted using the tidystm package by Mikael Poul Johannesson.2 Figure highlights nine topics with different expected proportions between male, AWS and non-AWS Labour MPs, with the error bars representing 95% confidence intervals. See Figure for a graph of all 66 topics.

\label{k0-topic-selected-bar-plot}Selected Topic Proportions

Selected Topic Proportions

\label{k0-topic-bar-plot}All Topic Proportions

All Topic Proportions

3.6.2 Word Occurences

The table below shows the twenty most common words in each topic, and the twenty words with the highest FREX score, a measure that uses a harmonic mean of word exclusivity and topic coherence (Airoldi & Bischof, 2016). We have named each topic based on the most common words and highest FREX score words in each topic.

Words in Topic
Topic Number Top Twenty Words Top Twenty FREX
  1. Employment & unions
rights, workers, law, human, civil, trade, union, protection, employers, act, employment, unions, safety, employees, work, service, staff, employer, legislation, protect tupe, blacklisting, acas, rights, gangmasters, civil, dispute, protections, unions, dismissal, servants, human, disputes, workers, employer, num, certification, employees, tuc, employers
  1. Legal system
cases, court, legal, case, justice, law, courts, evidence, lord, appeal, system, criminal, judicial, investigation, judge, aid, prosecution, circumstances, trial, lawyers judicial, attorney-general, court, prosecutor, judges, carlile, defendant, extradition, cps, judiciary, admissible, pre-charge, jury, solicitors, lawyers, solicitor, courts, lawyer, detention, judge
  1. Roads
road, planning, site, land, sites, car, vehicles, residents, roads, safety, use, driving, vehicle, park, development, traffic, drivers, area, cars, speed bikes, cyclists, pedestrians, gypsy, off-road, cycling, encampments, parking, highways, masts, drivers, belt, roads, highway, road, gypsies, vehicles, site, vehicle, bike
  1. Housing
housing, homes, social, affordable, property, home, properties, london, accommodation, building, private, houses, tenants, rent, need, council, landlords, sector, buy, people tenants, rent, landlords, rented, homelessness, rents, leaseholders, leasehold, tenancy, commonhold, hmos, housing, one-bedroom, homeless, properties, right-to-buy, affordable, sleepers, fulham, landlord
  1. Police, firefighters & prison
police, officers, crime, policing, service, fire, prison, home, force, chief, community, officer, staff, forces, neighbourhood, probation, prisons, safety, prisoners, resources policing, firefighters, constables, pcsos, probation, csos, prisons, fire, constable, hmic, constabulary, officers, police, prison, prisoners, reoffending, neighbourhood, metropolitan, fires, ipcc
  1. Northern Ireland
make, sure, progress, northern, decisions, ireland, difference, towards, future, process, contribution, statement, responsibilities, easier, responsibility, must, departmental, belfast, friday, choices sinn, fein, make, sure, belfast, northern, progress, ulster, difference, ireland, ruc, decisions, patten, dissident, departmental, taoiseach, antrim, imc, chastelain, dpps
  1. Committee
committee, report, review, commission, independent, government, select, process, evidence, inquiry, scrutiny, recommendations, role, board, set, work, reports, public, published, parliament committee’s, select, inquiry, scrutiny, recommendations, committee, committees, independent, recommendation, panel, pre-legislative, report, chairman, review, reviews, scrutinise, inquiries, conclusions, publication, findings
  1. Schools
schools, school, education, teachers, pupils, primary, children, standards, educational, special, secondary, parents, free, teacher, teaching, head, academies, academy, curriculum, good schools, teachers, pupils, academies, pupil, grammar, classroom, leas, school’s, academisation, school, teacher, bsf, academy, headteachers, ofsted, lea, literacy, curriculum, classrooms
  1. Energy & climate change
energy, climate, change, fuel, carbon, gas, power, emissions, waste, nuclear, prices, wind, green, environmental, electricity, oil, industry, efficiency, renewable, price energy, carbon, electricity, renewable, renewables, solar, ofgem, greenhouse, co2, ccs, feed-in, biofuels, microgeneration, fossil, sellafield, decarbonisation, chp, shale, mw, bnfl
  1. Defence
defence, forces, armed, afghanistan, service, military, personnel, army, security, troops, support, ministry, royal, veterans, british, force, capability, iraq, equipment, also armed, veterans, mod, regiment, legion, servicemen, reservists, helmand, battalion, ta, hms, gurkhas, regiments, marines, gurkha, fusiliers, ex-service, eurofighter, isaf, afghan
  1. Parliament
house, leader, motion, commons, therefore, parliament, petition, parliamentary, government, urge, present, signed, table, notes, library, behalf, remain, floor, westminster, request petitioners, declares, petition, house, motion, urges, commons, serjeant, recess, notes, leader, motions, lobbyist, thursday, early-day, e-petitions, house’s, tuesday, session, lobbying
  1. International politics
united, states, agreement, kingdom, foreign, treaty, council, security, us, nuclear, president, co-operation, convention, nations, national, policy, article, russia, international, position lisbon, ratification, treaty, non-proliferation, treaties, qmv, ratified, veto, gibraltar, ukraine, russia, agreement, protocol, states, united, ratify, russian, kingdom’s, hague, disarmament
  1. Ministers
secretary, state, statement, ministers, today, confirm, department, government’s, explain, yesterday, home, plans, announcement, government, welcome, chief, state’s, urgent, ministerial, announced secretary, state, state’s, confirm, ministers, yesterday, announcement, ministerial, explain, statement, expects, urgent, intends, assurances, yesterday’s, secretaries, secretary’s, update, leaked, cabinet
  1. Policy impact
made, clear, number, decision, impact, changes, recent, assessment, effect, level, discussions, likely, proposed, colleagues, potential, representations, implications, analysis, effects, result made, clear, decision, assessment, recent, changes, impact, representations, implications, effect, discussions, analysis, assess, implementation, estimate, level, number, negative, outcome, colleagues
  1. Gender
women, men, violence, equality, domestic, age, discrimination, women’s, equal, pay, woman, girls, gender, sexual, sex, female, gap, government, maternity, male women’s, gender, transgender, breastfeeding, refuges, women, abortions, fgm, shortlists, female, male, equality, girls, all-women, gay, equalities, lesbian, men, pregnancy, fawcett
  1. Regional development
new, development, future, programme, national, strategy, government, regional, key, plan, department, welcome, paper, set, ensure, commitment, support, improve, need, deliver strategy, regional, programme, projects, paper, plan, project, deliver, white, key, development, delivering, develop, priorities, partnership, improve, framework, new, priority, improving
  1. Communications
office, post, bank, banks, rural, offices, services, service, royal, banking, network, mail, closure, access, areas, broadband, card, account, staff, closures offices, mail, sub-postmasters, sub-post, superfast, post, postwatch, postcomm, consignia, broadband, rbs, office, banking, mail’s, bank, lloyds, ons, uso, branches, banks
  1. Immigration
british, uk, rules, home, immigration, citizens, asylum, identity, status, country, overseas, application, indicated, applications, apply, border, abroad, cards, migration, entry passports, nationality, dissent, immigration, passport, indicated, points-based, identity, asylum, nationals, visa, dependencies, migration, migrants, biometric, overseas, citizen, entry, abroad, monarch
  1. Health system
health, nhs, hospital, service, patients, services, mental, trust, staff, hospitals, care, trusts, patient, primary, waiting, doctors, nurses, e, gp, emergency in-patient, helier, nurses, chcs, nhs, ccgs, ccg, sha, hospital’s, hospital, fundholding, pct, hospitals, mental, gp, healthwatch, orthopaedic, walk-in, trusts, reconfiguration
  1. International development
international, countries, world, aid, development, government, developing, africa, global, uk, support, trade, poverty, country, india, assistance, un, need, also, nations zimbabwe, dfid, burma, congo, cdc, kenya, burmese, doha, uganda, mugabe, sub-saharan, g8, zimbabwean, dfid’s, gleneagles, african, sri, lanka, cancun, nigeria
  1. Benefits & disability
people, benefit, work, benefits, disabled, support, allowance, welfare, employment, disability, system, government, help, universal, credit, reform, get, vulnerable, plus, living incapacity, dla, esa, jobcentre, disabled, jobseeker’s, jsa, disability, allowance, dwp, claimants, atos, benefit, plus, claiming, pip, motability, benefits, deaf, bedroom
  1. Sport & culture
city, centre, town, sport, football, community, liverpool, sports, club, constituency, clubs, culture, london, great, facilities, one, bid, games, towns, regeneration football, olympic, museum, museums, stadium, athletes, cricket, paralympic, games, gospels, sports, club, sporting, fans, cup, rugby, arts, olympics, sport, galleries
  1. History
history, former, world, tribute, great, day, never, proud, first, remember, new, john, campaign, century, parliament, pay, also, war, today, sir maiden, miners, memorial, predecessors, hillsborough, tony, martin, james, john, william, andrew, margaret, anniversary, peter, alan, memories, fought, harold, churchill, edward
  1. Higher education & skills
education, skills, students, university, training, higher, young, universities, college, learning, science, apprenticeships, colleges, fees, student, funding, research, system, qualifications, courses universities, student, apprenticeship, fe, graduates, ema, graduate, students, colleges, diploma, apprenticeships, vocational, leitch, esol, qualifications, courses, undergraduate, university, tuition, sixth-form
  1. Concurring point
point, agree, country, making, makes, absolutely, whole, much, good, part, friend’s, entirely, completely, kind, sense, giving, rather, share, precisely, parts agree, absolutely, makes, friend’s, point, precisely, making, entirely, completely, kind, whole, sense, direction, mentions, refers, gentleman’s, describes, powerful, danger, exactly
  1. Pensions
scheme, pension, credit, pensions, insurance, schemes, pensioners, payments, compensation, fund, payment, money, financial, paid, savings, debt, retirement, government, pay, income pension, annuity, policyholders, annuities, auto-enrolment, insurance, retirement, loan, payments, payday, scheme, compensation, equitable, premiums, payment, pensions, means-testing, lenders, savers, pensioners
  1. Points of order
question, order, mr, put, speaker, deputy, point, grateful, read, agreed, record, time, minutes, may, call, standing, correct, apologise, madam, interventions speaker, mr, madam, question, forthwith, deputy, apologise, order, o’clock, read, minutes, adjourned, accordingly, interventions, hansard, tomorrow, grateful, misled, correct, courtesy
  1. Issues
important, issue, can, issues, take, ensure, hope, need, matter, consider, possible, place, also, concerns, deal, particular, course, taken, concern, raised issues, issue, important, concerns, consider, possible, discuss, concern, particular, matter, considering, carefully, assure, understand, extremely, raised, addressed, obviously, address, expressed
  1. Constituencies
many, constituency, constituents, problems, welcome, particularly, people, often, hard, face, others, feel, country, especially, worked, pay, concerned, represent, thousands, large many, constituents, problems, hard, mine, worked, difficulties, faced, represent, feel, constituencies, thousands, hundreds, face, greatly, often, constituency, especially, worried, experienced
  1. Ethnic groups & racism
action, taking, community, steps, taken, communities, take, actions, society, prevent, faith, groups, minority, church, black, ethnic, religious, freedom, race, diversity religion, faiths, sikh, steps, racial, faith, sikhs, religious, priests, synod, beliefs, church, racism, taking, action, ethnic, anglican, hate, clergy, hatred
  1. Amendments
clause, amendment, amendments, new, lords, section, 1, tabled, 2, clauses, line, 3, leave, act, shall, move, beg, 4, page, schedule insert, nos, subsection1, amendmenta, amendment, subsection5, 1a, schedule, amendmentsa, amendments, subsection2, subsection6, clause, tabled, paragrapha, subsection, subsection3, andc, paragraphb, clauses
  1. Reports
year, since, report, number, figures, official, march, april, published, 1997, figure, statistics, 15, 30, show, january, 2010, july, june, december vol, october, march, official, february, july, january, november, june, april, 2011, statistics, since, 2009, 2007, december, 2005, figures, 2013, figure
  1. People
people, want, get, one, go, can, think, see, need, know, say, things, much, like, good, going, problem, done, something, put things, get, something, go, lot, want, talking, thing, trying, talk, think, really, quite, bit, else, happen, away, getting, enough, idea
  1. Wales & Scotland
wales, scotland, scottish, england, welsh, assembly, parliament, devolution, uk, devolved, government, powers, kingdom, national, english, united, glasgow, executive, snp, edinburgh scotland, scottish, welsh, snp, scotland’s, cymru, barnett, plaid, perth, wishart, holyrood, perthshirepete, wales, snp’s, assembly, devolved, dundee, scots, devolution, calman
  1. Alcohol & tobacco
food, industry, alcohol, licensing, products, smoking, shops, shop, tobacco, advertising, health, standards, pub, pubs, high, buy, drinking, supermarkets, problem, retailers tobacco, pubs, gambling, betting, labelling, drinks, cigarettes, casinos, smokers, cigarette, groceries, lap-dancing, vending, drinkers, supermarkets, fluoride, smoking, pubcos, pub, retailers
  1. Place names
thank, south, constituency, north, excellent, join, congratulate, manchester, area, yorkshire, north-west, reply, visit, greater, visited, also, bristol, nottingham, giving, region thank, wrexham, reddish, tameside, congratulating, newport, yorkshire, stockport, blaenau, derbyshire, south, north-west, stoke-on-trent, denbighshire, denton, nottingham, bristol, welcoming, newingtonms, congratulations
  1. Budget
million, budget, year, billion, cuts, chancellor, spending, cut, increase, money, government, 1, funding, extra, next, investment, deficit, financial, crisis, growth deficit, obr, billion, spending, budget, real-terms, forecast, million, borrowing, cuts, gdp, chancellor, cut, 2.5, chancellor’s, forecasts, 2010-11, 1.2, 1.5, finances
  1. Tax
tax, pay, rate, income, wage, families, minimum, living, low, poverty, working, vat, increase, government, paid, national, paying, credits, average, poorest tax, millionaires, 50p, vat, taxes, credits, wage, taxation, avoidance, incomes, rate, zero-hours, wages, 45p, earning, revaluation, income, richest, earners, regressive
  1. Private companies
companies, company, market, financial, industry, competition, consumers, interest, consumer, assets, services, profits, markets, ownership, regulator, share, corporate, interests, customers, societies mutuals, shareholders, provident, company, companies, competition, profits, corporate, shares, company’s, societies, co-operative, fsa, co-operatives, profit, directors, rock, regulator, assets, asset
  1. Environment & fishing
environment, sea, fishing, marine, fisheries, industry, natural, fish, port, environmental, water, ports, rural, coastal, protection, conservation, fishermen, areas, management, area fishing, fisheries, fishermen, cod, seas, whitby, coastguard, broads, cfp, angling, seafarers, anglers, inshore, discards, mmo, under-10, sssis, dredging, cockle, aonbs
  1. Crime
crime, behaviour, victims, offence, criminal, serious, abuse, offences, antisocial, home, use, measures, drugs, drug, enforcement, offenders, problem, tackle, law, justice sentences, asbos, cannabis, antisocial, offences, offence, trafficking, gangs, behaviour, penalty, sentencing, sentence, theft, criminals, custodial, offending, knife, heroin, offenders, victim
  1. Bills
bill, legislation, act, new, powers, provisions, regulations, power, place, provision, duty, apply, statutory, necessary, allow, provide, set, already, introduce, require provisions, bill, bill’s, definition, legislation, regulations, statutory, passage, seeks, requirement, drafted, draft, statute, intention, safeguards, purpose, consult, legislative, amend, covered
  1. Children
children, child, parents, families, children’s, support, poverty, family, young, needs, parent, start, adoption, adults, vulnerable, early, contact, must, need, autism autism, csa, looked-after, adoptive, child, adopters, children’s, autistic, cafcass, nspcc, child’s, children, parent, dyslexia, adoption, kinship, childcare, intercountry, parents, lone
  1. Utilities & PFI
public, private, sector, money, costs, cost, risk, value, management, service, water, government, contracts, contract, system, audit, flood, systems, agency, taxpayer id, flood, nao, ofwat, public, contracts, private, auditor, purse, contractors, audit, pac, pfi, flooding, taxpayer, floods, contract, comptroller, tendering, defences
  1. Middle East
security, government, peace, war, foreign, people, iraq, terrorism, international, conflict, threat, support, must, un, military, syria, israel, resolution, terrorist, refugees syria, israel, palestinian, israeli, gaza, palestinians, syrian, saddam, arab, hamas, saudi, daesh, palestine, isil, israelis, hussein, lebanon, atrocities, assad, two-state
  1. Local authorities
local, authorities, council, authority, areas, government, funding, area, councils, communities, county, grant, planning, community, central, formula, borough, locally, level, resources local, authorities, councillors, councils, authority, unitary, county, formula, grant, lga, locally, localism, swindon, allocations, allocation, deprived, council, parish, authority’s, deprivation
  1. Elections
vote, political, parliament, electoral, election, elections, elected, parties, people, voting, referendum, democracy, register, system, registration, democratic, commission, party, votes, majority electoral, voters, turnout, voter, all-postal, votes, vote, voting, polling, first-past-the-post, av, referendums, elections, unelected, registration, ballot, candidates, electors, electorate, elected
  1. Debate
members, debate, speech, heard, today, hope, opportunity, speak, hear, chamber, great, wish, support, time, pleased, debates, sides, like, follow, subject debate, speech, members, debates, speeches, speak, heard, listened, sides, debating, hear, speaking, tonight, pleasure, chamber, thoughtful, listening, afternoon, queen’s, cross-party
  1. Transport
london, transport, rail, bus, services, line, network, travel, airport, train, air, service, passengers, trains, railway, station, east, capacity, passenger, heathrow rail, bus, passengers, trains, passenger, heathrow, railways, fares, freight, crossrail, hs2, high-speed, runway, electrification, airlines, gatwick, caa, baa, sra, thameslink
  1. Questions
whether, information, may, answer, asked, ask, questions, response, available, advice, received, data, know, press, written, letter, department, meeting, details, officials answer, information, questions, answers, data, written, details, letter, write, ask, officials, answered, asked, whether, informed, press, website, correspondence, response, requests
  1. Families
family, life, families, lives, constituent, death, home, people, told, case, one, man, died, lost, mrs, person, mother, day, marriage, suffered husband, mum, daughter, constituent, married, mrs, son, mother, marriage, died, father, wife, same-sex, death, loved, dad, suicide, funeral, bereaved, boy
  1. Health research
research, treatment, cancer, medical, disease, health, drugs, condition, can, use, drug, patients, screening, risk, also, conditions, evidence, group, diseases, diagnosis screening, asbestos, tissue, embryos, cancers, hepatitis, genetic, prostate, epilepsy, cloning, pleural, fertilisation, embryo, embryonic, ivf, anaemia, embryology, piercing, hfea, bowel
  1. Dispatch box
back, come, look, forward, bring, moment, coming, comes, side, later, brought, along, bringing, round, looking, box, see, putting, sit, dispatch come, back, look, moment, forward, dispatch, coming, comes, side, box, oh, surprise, bring, round, hoping, bringing, sooner, straight, along, sit
  1. Parties
government, labour, conservative, party, opposition, policy, previous, liberal, conservatives, government’s, support, election, tory, front, democrats, coalition, benches, policies, general, fact conservative, conservatives, liberal, democrats, lib, tory, democrat, benches, tories, opposition, manifesto, party’s, labour, benchers, dem, opposition’s, front-bench, party, spokesman, bench
  1. Statements
us, said, just, let, say, now, tell, says, yet, saying, told, know, going, nothing, wrong, even, wants, words, minister’s, today tell, says, let, wants, us, actually, saying, minister’s, telling, truth, wrong, wonder, thinks, nothing, promise, afraid, mistake, blame, admit, honest
  1. European Union
european, eu, europe, union, uk, countries, britain, trade, single, british, negotiations, market, economic, france, germany, country, leave, membership, referendum, world euro, ttip, brexit, accession, eu, currencies, cypriots, european, eurozone, europe, enlargement, pro-european, spain, currency, esm, france, greece, italy, brussels, isds
  1. Locations
member, west, east, north, birmingham, friends, st, spoke, hull, sheffield, talked, leeds, leicester, midlands, upon, newcastle, westmr, eastmr, northmr, southmr kingston, eastmr, bromley, chislehurstmr, holborn, dorsetmr, northmr, enfield, hull, southmr, chislehurst, stuart, ealing, rees-mogg, leicester, chingford, westmr, greenmr, southend, letwin
  1. Jobs & manufacturing
jobs, economy, economic, growth, industry, unemployment, investment, government, uk, manufacturing, future, sector, employment, country, job, long-term, steel, north-east, industries, recession steel, manufacturing, jobs, tata, economy, teesside, unemployment, recession, automotive, downturn, steelworkers, productivity, inward, growth, industries, recessions, nissan, economic, steelworks, double-dip
  1. Small business
business, small, businesses, regulation, rates, enterprise, government, finance, support, firms, help, innovation, measures, regulatory, smaller, large, lending, enterprises, burden, larger smes, medium-sized, businesses, business, enterprises, small, regulation, enterprise, commerce, entrepreneurs, tape, firms, lending, burdens, brs, start-up, start-ups, entrepreneurial, lend, smaller
  1. Agreement & disagreement
believe, however, one, might, accept, must, different, case, system, view, change, think, whether, position, argument, rather, simply, reason, basis, although accept, argument, principle, view, arguments, reason, might, argue, perfectly, suggest, balance, believe, suggesting, different, reasons, necessarily, sensible, disagree, argued, whatever
  1. Voluntary sector
work, people, young, support, help, can, working, organisations, role, voluntary, ensure, together, good, also, need, important, encourage, opportunities, experience, society voluntary, organisations, charities, volunteering, young, charity, youth, work, opportunities, helping, encourage, volunteers, encouraging, play, charitable, working, help, ways, valuable, together
  1. Comments
member, said, shall, mentioned, earlier, points, lady, comments, referred, learned, intervention, remarks, interesting, raised, pointed, perhaps, gave, say, refer, described comments, remarks, lady, interesting, points, happily, southwark, referred, bermondsey, referring, somerton, intervention, shall, intervened, mentioned, pointed, learned, earlier, gentlemen, rushcliffemr
  1. Social care
care, services, social, carers, people, need, service, needs, support, provision, older, provide, quality, home, centres, access, elderly, provided, providers, homes carers, hospices, dentists, dental, care, dementia, hospice, dentistry, respite, carer, advocacy, elderly, older, caring, palliative, milton, dentist, social, keynes, cared
  1. Time
years, time, last, two, one, first, now, three, past, week, months, next, ago, every, 10, five, four, weeks, days, six years, three, two, last, months, ago, past, time, four, week, weeks, six, five, first, next, days, 10, seven, half, now
  1. Media & animals
bbc, farmers, digital, television, internet, animals, animal, media, radio, dogs, licence, dog, news, ban, farming, welfare, hunting, fee, online, farm bbc, dogs, hunting, cull, bbc’s, badgers, badger, bovine, switchover, broadcasters, gm, fur, mink, poultry, circuses, analogue, hare, hounds, puppies, swine
  1. Other
given, can, aware, may, recently, across, welcome, fact, government, well, take, close, result, seeking, indeed, support, confident, responsible, know, including given, aware, can, recently, may, across, close, welcome, fact, confident, seeking, result, well, take, responsible, indeed, keep, regret, far, reconsider

3.6.3 Manual Validation

We have validated both the topics produced by the model and our labels of those topics to ensure the topics themselves are both interesting and relevant. Validation is particularly important in unsupervised models including STM (Grimmer & Stewart, 2013). Quinn, Monroe, Colaresi, Crespin, & Radev (2010) suggest that topics are valid if they correspond to external events. Figure shows the number of speeches by Labour MPs on the “Middle East” topic, with a spike in 2003 (at the start of the Iraq War), another spike in 2008 and 2009, as the bulk of British troops left Iraq, a small spike in 2011 coinciding with UK participation in NATO’s military intervention in Libya, and another spike resulting from debate in 2014–2016 over UK participation in military interventions in the Syrian Civil War.

Figure shows debate over the devolved authorities of Wales and Scotland peaking in 2014, to coincide with Scotland’s independence referendum. The post-2015 decline also likely stems from the SNP winning all but three seats in Scotland during the 2015 General Election. Figure shows the increase in debate over the European Union coinciding with the referendum on the UK’s member of the European Union.

\label{middle-east-plot-validity}Number of Speeches in "Middle East" Topic per Year

Number of Speeches in “Middle East” Topic per Year

\label{wales-scotland-validity}Number of Speeches in "Wales & Scotland" Topic per Year

Number of Speeches in “Wales & Scotland” Topic per Year

\label{eu-validity}Number of Speeches in "European Union" Topic per Year

Number of Speeches in “European Union” Topic per Year

4 Discussion

There do not appear to be substantial or meaningful differences in the speaking styles of female Labour MPs selected through all women shortlists when compared to their female colleagues selected through open shortlists using LIWC. This is possibly due to the speaking style dominant in British parliamentary debate, which is more formal than the speech used in most day-to-day conversation. LIWC was developed by American researchers, and the LIWC dictionary may not be able to capture stylistic differences between American and British English, and may not include words commonly used in formal British English speech, limiting its usefulness in the context of British political debate.

There is more distinction between AWS and non-AWS MPs in terms and topics. Naive Bayes classification was able to accurately determine the AWS status of female Labour MPs with slightly greater accuracy than it could distinguish between male and female Labour MPs (71.22% and 70.67%, respectively).

AWS MPs are far more likely to make reference to their constituency and constituents. In the debate between whether MPs should be “delegates” or “trustees” – the “mandate-independence controversy” outlined by Pitkin (1967) – the references to their constituents and constituencies suggests AWS MPs shy away from the Burkean concept of trusteeship and see themselves more as strict representatives of their constituents. In Andeweg & Thomassen’s (2005) typology of ex ante/ex post and above/below political representation, AWS MPs lean towards representation “from below”, although their selection process is ex ante/ex post. AWS MPs also use events and individuals in their constituency as examples when speaking on a given topic (see the Appendix for more examples).

AWS MPs refer to their constituents both specifically and in the abstract, particularly when criticising government policy. For example, in debate on 4th March 2015, Gemma Doyle, than the Labour MP for West Dunbartonshire (elected on an AWS in 2010), when asked if she would give way to Conservative MP Stephen Mosley, responded:

No, I will not [give way], because my constituents want me to make these points, not to give more time to Conservative Members.

On 2nd June 2010, during debate on Israel-Palestine, Valerie Vaz, MP for Walsall South, also used the views of her constituents to support her position:

My constituents want more than pressure. Will the Foreign Secretary come back to the House and report on a timetable for the discussions on a diplomatic solution, just as we did on Ireland?

On 4th April 2001, Betty Williams, member for Conwy from 1997–2010, raised the case of a wilderness guide in her constituency unable to access parts of the countryside due to foot and mouth disease:

Is my right hon. Friend aware that there is continuing concern about the limited access to the countryside and crags of north Wales? May I draw his attention to the circumstances of my constituent, Ric Potter? Like many others, he has had to travel to Scotland, where there is greater access. Will my right hon. Friend help us to enable people such as Ric Potter to find work in outdoor pursuits?

5 Appendix

5.1 Gender effect estimates

Estimate effects of different topics, using only gender.

Topic Estimates
Estimate Standard Error t value Pr(>|t|)
Topic 1 – Employment & unions
Intercept 0.0120558 0.0001226 98.2982598 < 0.001 ***
Female -0.0009773 0.0002202 -4.4373689 < 0.001 ***
Topic 2 – Legal system
Intercept 0.0167081 0.0001695 98.5906502 < 0.001 ***
Female 0.0001904 0.0002876 0.6621819 0.51
Topic 3 – Roads
Intercept 0.0116843 0.0001466 79.6825193 < 0.001 ***
Female -0.0018067 0.0002565 -7.0437191 < 0.001 ***
Topic 4 – Housing
Intercept 0.0112593 0.0001766 63.7601028 < 0.001 ***
Female 0.0054862 0.0002889 18.9886898 < 0.001 ***
Topic 5 – Police, firefighters & prison
Intercept 0.0140240 0.0001728 81.1565375 < 0.001 ***
Female 0.0008923 0.0002990 2.9842081 0.003 **
Topic 6 – Northern Ireland
Intercept 0.0089596 0.0000453 197.9376090 < 0.001 ***
Female -0.0002230 0.0000831 -2.6833803 0.007 **
Topic 7 – Committee
Intercept 0.0213134 0.0001512 140.9203444 < 0.001 ***
Female -0.0015285 0.0002328 -6.5672623 < 0.001 ***
Topic 8 – Schools
Intercept 0.0147374 0.0001952 75.5156362 < 0.001 ***
Female 0.0009837 0.0003469 2.8358887 0.005 **
Topic 9 – Energy & climate change
Intercept 0.0170306 0.0002011 84.6909082 < 0.001 ***
Female -0.0026726 0.0003639 -7.3438573 < 0.001 ***
Topic 10 – Defence
Intercept 0.0157848 0.0001844 85.5795520 < 0.001 ***
Female -0.0061179 0.0003260 -18.7662038 < 0.001 ***
Topic 11 – Parliament
Intercept 0.0119034 0.0000793 150.1066439 < 0.001 ***
Female -0.0019536 0.0001431 -13.6535321 < 0.001 ***
Topic 12 – International politics
Intercept 0.0125950 0.0001233 102.1756834 < 0.001 ***
Female -0.0050772 0.0002152 -23.5953423 < 0.001 ***
Topic 13 – Ministers
Intercept 0.0167095 0.0001075 155.4918804 < 0.001 ***
Female 0.0011453 0.0001909 5.9990245 < 0.001 ***
Topic 14 – Policy impact
Intercept 0.0115405 0.0000467 247.1534072 < 0.001 ***
Female 0.0009546 0.0000873 10.9321624 < 0.001 ***
Topic 15 – Gender
Intercept 0.0048719 0.0001169 41.6786074 < 0.001 ***
Female 0.0121878 0.0002402 50.7432137 < 0.001 ***
Topic 16 – Regional development
Intercept 0.0230320 0.0001267 181.8358615 < 0.001 ***
Female 0.0026217 0.0002534 10.3464415 < 0.001 ***
Topic 17 – Communications
Intercept 0.0097486 0.0001203 81.0660742 < 0.001 ***
Female -0.0009969 0.0002068 -4.8214598 < 0.001 ***
Topic 18 – Immigration
Intercept 0.0087113 0.0000879 99.0798879 < 0.001 ***
Female -0.0000399 0.0001607 -0.2483047 0.80
Topic 19 – Health system
Intercept 0.0161813 0.0001963 82.4182372 < 0.001 ***
Female 0.0079699 0.0003606 22.0990638 < 0.001 ***
Topic 20 – International development
Intercept 0.0160467 0.0001888 84.9712499 < 0.001 ***
Female -0.0020680 0.0003324 -6.2206476 < 0.001 ***
Topic 21 – Benefits & disability
Intercept 0.0120721 0.0001443 83.6430500 < 0.001 ***
Female 0.0080950 0.0002813 28.7771977 < 0.001 ***
Topic 22 – Sport & culture
Intercept 0.0127416 0.0001522 83.7277733 < 0.001 ***
Female -0.0003740 0.0002616 -1.4293244 0.15
Topic 23 – History
Intercept 0.0137582 0.0001119 122.9285059 < 0.001 ***
Female -0.0046789 0.0001838 -25.4564522 < 0.001 ***
Topic 24 – Higher education & skills
Intercept 0.0143325 0.0001639 87.4390370 < 0.001 ***
Female -0.0004525 0.0002994 -1.5114308 0.13
Topic 25 – Concurring point
Intercept 0.0155206 0.0000466 332.8280261 < 0.001 ***
Female -0.0026311 0.0000750 -35.0700318 < 0.001 ***
Topic 26 – Pensions
Intercept 0.0147001 0.0001701 86.4362686 < 0.001 ***
Female 0.0019916 0.0002808 7.0937838 < 0.001 ***
Topic 27 – Points of order
Intercept 0.0177899 0.0001304 136.4542143 < 0.001 ***
Female -0.0054036 0.0002161 -25.0001241 < 0.001 ***
Topic 28 – Issues
Intercept 0.0345024 0.0000991 348.2518448 < 0.001 ***
Female 0.0006762 0.0001717 3.9389187 < 0.001 ***
Topic 29 – Constituencies
Intercept 0.0131792 0.0000538 245.0529092 < 0.001 ***
Female 0.0023290 0.0001065 21.8782635 < 0.001 ***
Topic 30 – Ethnic groups & racism
Intercept 0.0085781 0.0000726 118.0996216 < 0.001 ***
Female 0.0019576 0.0001366 14.3255282 < 0.001 ***
Topic 31 – Amendments
Intercept 0.0150300 0.0001583 94.9241895 < 0.001 ***
Female -0.0028644 0.0002684 -10.6722265 < 0.001 ***
Topic 32 – Reports
Intercept 0.0169717 0.0001105 153.6020779 < 0.001 ***
Female 0.0012798 0.0001857 6.8932095 < 0.001 ***
Topic 33 – People
Intercept 0.0377448 0.0001208 312.3638134 < 0.001 ***
Female -0.0014525 0.0002100 -6.9162948 < 0.001 ***
Topic 34 – Wales & Scotland
Intercept 0.0135395 0.0001542 87.8038603 < 0.001 ***
Female -0.0031719 0.0002505 -12.6629961 < 0.001 ***
Topic 35 – Alcohol & tobacco
Intercept 0.0108567 0.0001497 72.5174053 < 0.001 ***
Female 0.0005344 0.0002870 1.8623926 0.063
Topic 36 – Place names
Intercept 0.0083655 0.0000671 124.7427562 < 0.001 ***
Female 0.0007984 0.0001241 6.4322198 < 0.001 ***
Topic 37 – Budget
Intercept 0.0246516 0.0001794 137.3766916 < 0.001 ***
Female -0.0003427 0.0002989 -1.1463241 0.25
Topic 38 – Tax
Intercept 0.0193076 0.0001905 101.3312585 < 0.001 ***
Female 0.0030527 0.0003316 9.2047863 < 0.001 ***
Topic 39 – Private companies
Intercept 0.0123859 0.0001200 103.2265704 < 0.001 ***
Female -0.0009558 0.0002224 -4.2971025 < 0.001 ***
Topic 40 – Environment & fishing
Intercept 0.0094747 0.0001433 66.1392213 < 0.001 ***
Female -0.0024778 0.0002424 -10.2204118 < 0.001 ***
Topic 41 – Crime
Intercept 0.0141399 0.0001719 82.2729667 < 0.001 ***
Female 0.0052400 0.0003121 16.7910290 < 0.001 ***
Topic 42 – Bills
Intercept 0.0244361 0.0001510 161.8726636 < 0.001 ***
Female -0.0012074 0.0002583 -4.6746393 < 0.001 ***
Topic 43 – Children
Intercept 0.0076816 0.0001224 62.7703411 < 0.001 ***
Female 0.0094493 0.0002446 38.6253202 < 0.001 ***
Topic 44 – Utilities & PFI
Intercept 0.0123660 0.0000838 147.5182406 < 0.001 ***
Female -0.0001332 0.0001597 -0.8336308 0.40
Topic 45 – Middle East
Intercept 0.0174587 0.0002090 83.5492976 < 0.001 ***
Female -0.0020814 0.0003627 -5.7381267 < 0.001 ***
Topic 46 – Local authorities
Intercept 0.0179820 0.0001455 123.5806929 < 0.001 ***
Female 0.0015599 0.0002882 5.4129075 < 0.001 ***
Topic 47 – Elections
Intercept 0.0181818 0.0001562 116.4335256 < 0.001 ***
Female -0.0075881 0.0002715 -27.9527608 < 0.001 ***
Topic 48 – Debate
Intercept 0.0180195 0.0000686 262.8627614 < 0.001 ***
Female -0.0018249 0.0001230 -14.8305990 < 0.001 ***
Topic 49 – Transport
Intercept 0.0163769 0.0001851 88.4732521 < 0.001 ***
Female -0.0002980 0.0003461 -0.8609762 0.39
Topic 50 – Questions
Intercept 0.0161649 0.0000756 213.7358034 < 0.001 ***
Female 0.0001674 0.0001300 1.2874523 0.20
Topic 51 – Families
Intercept 0.0101121 0.0001161 87.1131980 < 0.001 ***
Female 0.0044915 0.0002511 17.8908688 < 0.001 ***
Topic 52 – Health research
Intercept 0.0087860 0.0001605 54.7287657 < 0.001 ***
Female 0.0050139 0.0002941 17.0467226 < 0.001 ***
Topic 53 – Dispatch box
Intercept 0.0075484 0.0000252 299.2492191 < 0.001 ***
Female -0.0010064 0.0000408 -24.6610302 < 0.001 ***
Topic 54 – Parties
Intercept 0.0248256 0.0001508 164.6595307 < 0.001 ***
Female -0.0062193 0.0002485 -25.0299952 < 0.001 ***
Topic 55 – Statements
Intercept 0.0211080 0.0000663 318.4773224 < 0.001 ***
Female -0.0025222 0.0001204 -20.9423749 < 0.001 ***
Topic 56 – European Union
Intercept 0.0163672 0.0001678 97.5490397 < 0.001 ***
Female -0.0044312 0.0002913 -15.2095098 < 0.001 ***
Topic 57 – Locations
Intercept 0.0100684 0.0001063 94.7537127 < 0.001 ***
Female -0.0008448 0.0001917 -4.4060051 < 0.001 ***
Topic 58 – Jobs & manufacturing
Intercept 0.0176038 0.0001734 101.5152313 < 0.001 ***
Female 0.0002192 0.0003130 0.7001742 0.48
Topic 59 – Small business
Intercept 0.0070549 0.0000703 100.3910783 < 0.001 ***
Female -0.0000223 0.0001173 -0.1899731 0.85
Topic 60 – Agreement & disagreement
Intercept 0.0328360 0.0001081 303.8892524 < 0.001 ***
Female -0.0102175 0.0001822 -56.0650145 < 0.001 ***
Topic 61 – Voluntary sector
Intercept 0.0187333 0.0001139 164.5326954 < 0.001 ***
Female 0.0075510 0.0002285 33.0471024 < 0.001 ***
Topic 62 – Comments
Intercept 0.0152788 0.0000597 255.8200094 < 0.001 ***
Female -0.0036510 0.0001006 -36.3055228 < 0.001 ***
Topic 63 – Social care
Intercept 0.0090885 0.0001331 68.2815486 < 0.001 ***
Female 0.0080707 0.0002302 35.0611376 < 0.001 ***
Topic 64 – Time
Intercept 0.0213923 0.0000679 315.0245762 < 0.001 ***
Female -0.0017936 0.0001260 -14.2325407 < 0.001 ***
Topic 65 – Media & animals
Intercept 0.0121559 0.0001645 73.9171975 < 0.001 ***
Female -0.0030924 0.0002736 -11.3010314 < 0.001 ***
Topic 66 – Other
Intercept 0.0038288 0.0000121 316.8868169 < 0.001 ***
Female 0.0002873 0.0000199 14.4065798 < 0.001 ***

5.2 \(\theta\) distribution

Figure shows the distribution of \(\theta\) scores used to assign overall topics to individual speeches in Table 10, per topic.

\label{k0-theta-boxplot}Theta Values in Topic Assignment

Theta Values in Topic Assignment

5.3 AWS References to Constituents in Context

A random selection of 2% of all references to “my constituency”, “my constituent” and “my constituents”, by AWS MPs, in context.

A random sample of KWIC’s
Pre Keyword Post
. I was briefed by a vehicle hire company in my constituency called Reflex and, quite frankly, if such banking
150 per cent. two years ago. Another of my constituents has advised me of an application for an 85 per
already begun, for example just over the border from my constituency in the constituency of my hon. Friend the Member
in which Cornish children study. Three secondary schools in my constituency will be located on the same site, and one
Manchester has been doing a major infrastructure project, and my constituents are at the end of their tether about the lack
patient at the BRI, and Airedale hospital is in my constituency . The hon. Member for South Cambridgeshire Mr.
, but the reality is there to be seen in my constituency . On Saturday I met a delegation of workers from
to use their abilities and develop their talents. In my constituency , 366 young people who have been unemployed for more
I believe that the most effective electoral registration officer in my constituency is mum. It is mum who fills in the
can arise from defective gas appliances, because two of my constituents , young students in their 20s, died from carbon
£ 3.6 million. Some 9% of people in my constituency are hard-working, entrepreneurial self-employed people, and today is
my right hon. Friend congratulate Aldercar community school in my constituency and its staff and pupils? The percentage of pupils
", One particular concern for many of my constituents is bus fares. As I have said, my
, Jobs and employment are the biggest issue in my constituency and the latest figures now show that just under 2,000
otherwise reach. The Psychiatric Rehabilitation Association is based in my constituency and was set up in 1959-it is no coincidence that
financial inclusion fund. Where would the Minister suggest that my constituents who are struggling with debt and excessive and escalating charges
and without the full participation of the British people, my constituents and the country will never forgive them.
. There is an additional problem that is relevant to my constituency . It contains a large outdoor venue called the National
if they continue to propose new services that, in my constituents ’ view, favour the administration of the hospital or
in red tape. That will be a turn-off. My constituency and the town in which it is situated has a
With my right hon. Friend’s local knowledge of my constituency , she will know that many of my constituents are
", to close a wide range of services at my constituency’s local hospital, St Helier. Most of the controversy
I am extremely worried for my constituents in Ashton-under-Lyne, Droylsden and Failsworth, and for people
One of the shortlisted sites is at Barnard Castle in my constituency , and that would produce 1,000 jobs.
making ends meet has been raised with me repeatedly by my constituents , including Graeme McGrory, who cares for his partner
One piece of transport infrastructure that my constituency and that of the hon. Member for Buckingham John
A director of Sirus Automotive who lives in my constituency would like to take on apprentices, but he has
" Three people who know that better than most are my constituents Mark, Joanne and Ben King. In 2011,
There are 3,540 women affected by the changes in my constituency . Does my hon. Friend agree that the 1995
have been down in the detail of rail provision in my constituency , but these are important matters for many of those
just a few examples of the work being done in my constituency . I recently had the privilege of accompanying the Gateshead
, but that does not help the large number of my constituents who have lost some, if not all, of
was the only mainstream candidate in the general election in my constituency who did not have their picture taken while pointing to
was not even in the mortgage application, NatWest told my constituents that it was in the process of adding it.
is a measure of the Government’s achievement that people in my constituency and elsewhere in Northamptonshire can look forward to a secure
clothing company announced the closure of two more factories in my constituency and the neighbouring Erewash constituency. A huge number of
my primary care trust in north-east Derbyshire and dentists in my constituency to find a local solution. These reforms coincide with
Cross, just a few miles up the road from my constituency . That pipe manufacturing works has been taken over by
go ahead. There is huge concern about this in my constituency and across the north. Was the Prime Minister told
backgrounds, including poor backgrounds, and is representative of my constituency . That is the sort of school that Labour Members
are subject to a TPIM. This information would let my constituents know whether potential terrorism suspects had returned to London.
. Gentleman for his generosity. Is he aware that my constituency is probably the one with the highest number of gas
because I have had direct experience of the issue in my constituency . A woman came over here as the wife of
. Let us take the feed-in tariff fiasco. In my constituency alone, we are losing many jobs, because a
What practical advice can the Secretary of State give to my constituents , as some 3,000 householders in my constituency face a
sport, that this is good enough for kids in my constituency ?
a fair deal on jobs, getting young people in my constituency and others involved in working our way out of the
argument is best explained by reference to the case of my constituent , Neil Kenny, who raised his concerns about the
to LEAs give rise to some questions, including in my constituency from Unison, which is concerned that LEAs might use
Such travel will be available to all 17,600 pensioners in my constituency . , In February I visited
", What point is there in forcing my constituent who is a single dad who has his two children
replies, perhaps he can respond to the questions that my constituent has raised. What is she to do? She
ask my hon. Friend to offer an undertaking to my constituents in Mitcham and Morden that an option appraisal of intermediate
he would be interested to hear the Minister’s response to my constituent Maureen Davenport. The Minister said that the maximum state
in child benefit, which will help 13,800 families in my constituency . My real reason for tabling the question is to
Finchley and Golders Green Mike Freer), many of my constituents killed by lorries have died at junctions, including some
Hall the plight of former United Engineering Forgings workers in my constituency who will not receive the returns from their final salary
London has had Oyster cards for nine years, but my constituents are still waiting. Although Transport for Greater Manchester is
again have a university. However, Nene college in my constituency hopes to change all that, and I support strongly
Enforcement Campaign-in Cardiff, and particularly to the work of my constituent , Professor John Shepherd, who works in the dental
and assets than non-disabled people. In London, where my constituency and the constituency of my hon. Friend the Member
in particular from the circumstances of students in Northampton. My constituency contains both a higher education and a further education college
the marine Bill on the grounds of its irrelevance to my constituents , because, like the hon. Lady, I
deepest concern for the families involved, especially given that my constituency neighbours that of my hon. Friend the Member for
services can expand on the slow line so that all my constituents benefit from the west coast main line upgrade?
rehabilitation. , The people of my constituency have been horrified by those cases, and it is
Labour Government we have achieved a tremendous amount. In my constituency the number of people claiming jobseeker’s allowance has almost halved
they complain? Where will the local accountability go? My constituents very much value the highly accessible local service that they
", Since helping the Jarrow marchers, my constituency has continued to welcome people from throughout the UK,
and not-for-profit groups, of which there are many in my constituency , doing immensely valuable work. They all too often
as soon as possible. Indeed, for some of my constituents , reform is already coming too late.
bus travel in Wales. I have met pensioners in my constituency who say that it has transformed their lives. As
and Sir Malcolm Thornton. All have represented part of my constituency and all left this House on 20 April or 1
Ports is the operator at the port of Immingham in my constituency . The companies there firmly believe that they have paid
Conservative-controlled Bradford city council excluded the wonderful Ilkley lido in my constituency from the free swimming initiative for young people and pensioners
for my hon. Friend’s reply, and many of my constituents who have come across the benefit integrity project will be
Tero was not properly treated and offer the apology that my constituent deserves.
about their corporate social responsibilities. For the sake of my constituents in Mitcham, Morden and Colliers Wood who want something
change in the law. Regrettably, not only in my constituency but in many northern towns and cities, I see
on an issue that has been of great concern to my constituents . While I appreciate the cross-party consensus that exists on
In my constituency of West Lancashire, the national lottery has supported 266
to meet the skills gap in engineering and construction in my constituency . , When I talk to
sat with the parents of the two children who were my constituents , as has Ken Livingstone, who made a private
who have been strongly encouraged to save The consultation in my constituency on the pensioners tax credit was extremely successful. The
Government for investing in the city of Bradford, helping my constituents to realise their potential. But in reality little has
visited Dot To Dot, a community arts project in my constituency . It has a good record of involving the community
one regret the fact that Westminster, which covers half my constituency , has so far concentrated CCTV bids-I am sure with
also significant gaps in the Bill. One example from my constituency concerns a community hydro project in Saddleworth that might not
hon. Friend for that reply, but most of my constituents probably do not know what a low carbon transition plan
has provided opportunities where there were none before. In my constituency , there have been far more opportunities in the past
to find examples of such practices. Another case in my constituency , with which I am dealing, involves elderly victims
. , The credit union in my constituency is fragile, because it serves an area in which
certainly applies to me because the acute trust that covers my constituents , who desperately need care, has the mother and
reveal a trend, and I see it happening in my constituency . It is a demonstrable fact that the polarisation between
My constituent , John Warren, has specifically asked me to raise
, Bridges Project in Musselburgh in my constituency does a brilliant job in supporting young people. A
Spowart, a small firm of legal aid solicitors in my constituency . Solicitors at the firm are paid generally between £
, both as a national concern and as it affects my constituency . I am grateful to my hon. Friend for
, nor, sadly, are far too many of my constituents .
My constituents in Hull are baffled by the Government’s approach. At
issue and go after these criminals who are preying on my constituents ?
even begin for another 12 months. Young people in my constituency should not have to spend another year on the dole
with the nutrition they need outside term time. In my constituency , several schools run summer programmes funded through the pupil
takes umbrage at being forced to do repairs-as some of my constituents , sadly, know to their cost.
", I recently visited a care home in my constituency that is provided by a small charity and is rated
House and members of the armed forces, such as my constituent , 19-year-old Private James Kenny of C company, 3rd
as out to Kent. There are seven stations in my constituency : Hither Green, Blackheath, Lee, Grove Park
Can my right hon. Friend give any assurance to my constituent , Mr. Peter Dyson, who has written to
Commons Library to conduct an analysis of the impact in my constituency . I discovered that 4,300 women and 3,800 men would
100 days of the new Parliament? Many businesses in my constituency are struggling significantly and would undoubtedly welcome a period of
in 1992, as the Member for Woolwich, before my constituency was formed for the 1997 election. John Austin is
were building up and seemed to take action only once my constituents had suffered a very high level of nuisance and there
that further education institutions, such as Blackburn College in my constituency , will not receive a real-terms funding cut as a
", On a more serious note, my constituency is home to manufacturers varying from Corus to Cadbury,
costs and cuts to working tax credits, families in my constituency will be worse off. I will not vote in
be warm. It paid for basics like that in my constituency . I will not revisit the pain of tuition fees
is a national issue. The 900 steel workers in my constituency whose jobs are on the line expect him to guarantee
to begin by speaking about the NHS as experienced by my constituents . Getting an appointment to see a GP can be
I was struck by what one of my constituents said last weekend, which was that the attacks that
", On 18 February, Llandudno in my constituency hosted the first North Wales criminal justice board conference.
my hon. Friend foresee for the young people in my constituency if they are to suffer possible cuts alongside that idiosyncratic
busways and widen the M1. Is he aware that my constituency will have the new Translink guided busway by 2008 due
" Last week, I hosted a jobs fair in my constituency , as have many hon. Members on both sides
in the south-east will be dealt with in Parliament? My constituents want to know where we are going and what the
him to visit the brand-new children’s centre in Elland in my constituency , which is due to open in January, and
realities for people affected by this situation. One of my constituents is stuck out in Saudi Arabia. His work has
the past few days. When the problems started in my constituency on Monday night, we saw copycat criminality, mindless
those branches, in Catford and Blackheath, are in my constituency and two others, in Lewisham and Greenwich, are
My constituent , Richard Belmar, has now spent nearly three years
Postwatch because I am unhappy about the consultation process in my constituency . I fully accept many of my hon. Friend’s
area of Keighley last Friday and talking to many of my constituents and taking on board many of their anxieties. On
of the major issues raised with me by carers in my constituency . We must take such issues on board. 
that the voucher company Farepak, which is based in my constituency , collapsed this week, robbing thousands of people on
scientific reports recommend restricted phone use by younger children. My constituents do not believe that such recommendations tally with the telecommunications
. Mullin). This is a big issue in my constituency , where inappropriate development on garden sites is taking place
scrutiny process, but it is impossible for me, my constituents or councillors of any party not involved in that enterprise
", At the time, I was consulting my constituents about their attitudes to crime and antisocial behaviour, and
you prove it? , My constituency is served by two hospitals: Dewsbury and District hospital
% reduction. What reassurances can the Minister give to my constituents and firefighters that those latest cuts will not jeopardise or
. , Horwich visiting service in my constituency has lost funding and can no longer employ its part-time
I have spoken to many businesses in my constituency . Will the hon. Gentleman concede that the Government’s
prevent businesses from going into administration, as many in my constituency are likely to do. Finally, the local authority
I do not know whether my experience in my constituency has been exactly the same as that of my right
? , Many SMEs operate in my constituency , and I want to ensure that the skills base
that population live in Salford, the local authority for my constituency . , In last year’s debate
It is an issue that has been simmering away in my constituency and recently the rumours have turned to reality as the
of the parenting lessons that go on in schools in my constituency to great effect. The hon. Gentleman ignores those
a distraught couple who run a hedgehog rescue centre in my constituency . They are currently nursing back to health a hedgehog
people to think that that was the total sum of my constituency . It is an extremely nice place to spend Christmas
transparency about the impact. , My constituents are also anxious about the Government’s proposals to allow fracking
some of its provisions will have on vulnerable people in my constituency . , I shall first raise
key elements of creative business growth. Creative businesses in my constituency and in a large area to the west of London
In Pembrokeshire we have two oil refineries, one in my constituency . They were both affected by the blockades in September
thank the Minister for his reply. Head teachers in my constituency are concerned that Government have still not come forward with
the work of local authorities in my area. In my constituency , there are no high profile arts venues that hit
many of the early asbestosis claims from Hebden Bridge in my constituency might not have succeeded under the proposed 75 per cent
job first.  , My constituency is pronounced  Erreywash , not 
that is not regulated properly, with the result that my constituents , who have small sums of money available to invest
a picture of the winning design, but people in my constituency have seen many pictures before. I want work to
hour. I have written to all the headteachers in my constituency over the last few weeks, and they tell me
this debate falls on an anniversary well worth remembering for my constituents . It is 20 years to the month that post-war
people of the east end, including the people of my constituency , talk to me about how excited they still are
I recently visited Bishop Barrington school in my constituency , which has got a new science lab and sports
the extent of the disruption and the problems caused for my constituents ? I would be happy to do that. 
increase in the number of new homes being built in my constituency over the past 10 years or so. For the
junior doctors who are the problem, but him? My constituents-hundreds of whom have written to me-overwhelmingly feel that he has
, , I do not think my constituents knew whether to laugh or cry.
about to be built in Walkden in the centre of my constituency . The new local improvement finance trust-LIFT-centre will include GP
is higher, and the dole queue is lengthening. My constituents are only too well aware of the exploitative practices of
" I am fortunate in having a research centre in my constituency at the university of Durham, which concentrates on enabling
is talking about the wrong hospital, which many of my constituents will find most amusing.
of the Land Registry would be bad not just for my constituents but for the public as a whole. The revenue
The food banks in my constituency , which currently number at least six, tell me
of those issues. , In my constituency , the credit union benefits from capital and revenue from
children. I am indebted to a law company in my constituency called Just for Kids Law, which has raised with
hope they are not giving false hope to many of my constituents . Will they just admit that they have made a
I have a range of energy-intensive industries in my constituency , including steel, glass, paper and the entire
the save Lewisham hospital campaign. But for now, my constituents still face the prospect of seriously downgraded services at their
from and bugbear for my constituents. On behalf of my constituents and their families, I very much look forward to
", helped motorists and the hard-pressed hauliers in my constituency-or they could have looked at jobs for young people.
Staff at Trinity, Bluecoat and Fernwood schools in my constituency are desperate for extra investment in their buildings. Will
The point about geography is critical in Cumbria, where my constituency is. Under the proposals, we will end up
will affect disabled youngsters. The What? centre in my constituency , which gives counselling to all youngsters, still does
closure of the offices is having a direct impact on my constituency . Walsall faces the closure of its HMRC office,
. , Frustration is evident among my constituents : for many years, they have felt marginalised and
, larger numbers of people are choosing to live in my constituency but work in London. If we are to take
ethnic minority children, of whom there are many in my constituency . , We have dealt a
single parents in the country-I will return to that point-and my constituents think that the measure is unfair. How people in
should not come back from our holidays to find that my constituents , and those of my neighbours, have lost their
their area; I fully intend to do so in my constituency . , We also need better
too much movement. I want Airedale general hospital in my constituency not just to survive, but to prosper. It
", During the summer and autumn months, my constituents and those of many other hon. Members were affected
put a human face on many of the difficulties that my constituents experience. , In Newham,
Parent Action Network, which has its national headquarters in my constituency . It has just received nearly £ 400,000 in lottery
sector. On Friday, an independent community pharmacist in my constituency told me that he estimated that the Government cuts would
it becomes an empty gesture. A community group in my constituency is setting up a community development trust, and it
since June and doubled since 2006. Young people in my constituency have been particularly badly hit, with a 288%
police get back to strength to defend the people in my constituency of Mitcham and Morden?
to address have been influenced by what has happened in my constituency in the past 10 days as a series of incidents
, including those of Allied Steel and Wire’s pensioners in my constituency ? They took the case to court through the unions
Indeed, it is a stealth cut. In my constituency , the Tories will have to make stealth cuts such
communities across the UK. I understand the concerns of my constituents . I understand that when a family from a different
a vested interest in ensuring the safety and security of my constituency , which in the past has been a military target
infrastructure project is a massive economic opportunity for Wales and my constituency in particular. Will the Minister assure the House that
Nottingham that stands to lose most is the Meadows in my constituency . Before the last election, the Meadows, one
am here this afternoon specifically to represent the concerns of my constituents who are trade union members in Parliament, as they
. Nothing could be further from the truth, as my constituency exemplifies. As I have already said, I represent
making are the very ones that have been made by my constituents , by the constituents of my hon. Friends and-I
, but wanted to take the opportunity to read out my constituent’s comments so that Ministers understand the worry and concern.
firm of Hickman and Rose, which is based in my constituency ? She was due to speak at a conference organised
Majesty’s Opposition. That public money could be used for my constituent Grace Ryder, aged 9, who was recently diagnosed
changes that will affect 650 families and 1,500 children in my constituency . , These are ideologically driven
deal more about the birdlife in both estuaries that surround my constituency . , The Bill establishes a
My constituent , the wonderful campaigner Marie Lyons, has doggedly pursued
  vote for their Muslim brother . My constituents were told that that was their religious duty. When
. It will bring huge benefits to many families in my constituency who are on low or not very generous incomes.
anywhere. , The diversity of my constituency is one of the reasons why it is the best
c The NHS in my constituency has moved beyond special measures into the success regime.
invited my right hon. and learned Friend to meet my constituents to hear what they think about our local NHS.
fleeing Ebola-affected countries are not left destitute and homeless? My constituents , Mr and Mrs Mahmood, have been working in
pension credit, but I wondered whether Ministers could give my constituent and me advice on whether the notional sum tied up
first home. There are so many young people in my constituency who see homes priced out of their reach and for
There are also problems for low-income families, such as my constituent on Colleymoor Leys lane who says:
term. I know from the experience of businesses in my constituency and in the surrounding west midlands area that New Street
that he needs those, but he failed to tell my constituents watching yesterday that a 1p cut in duty will not
average, which show that over a fifth-22% in my constituency-of people who resort to food banks for an emergency food

References

Airoldi, E. M., & Bischof, J. M. (2016). Improving and Evaluating Topic Models and Other Models of Text. Journal of the American Statistical Association, 111(516), 1381–1403. https://doi.org/10.1080/01621459.2015.1051182

Andeweg, R. B., & Thomassen, J. J. (2005). Modes of Political Representation: Toward a New Typology. Legislative Studies Quarterly, 30(4), 507–528. https://doi.org/10.3162/036298005X201653

Arora, S., Ge, R., Halpern, Y., Mimno, D., Moitra, A., Sontag, D., … Zhu, M. (2013). A Practical Algorithm for Topic Modeling with Provable Guarantees. In S. Dasgupta & D. McAllester (Eds.), Proceedings of the 30th International Conference on Machine Learning (Vol. 28, pp. 280–288). Atlanta, Georgia, USA: PMLR. Retrieved from http://proceedings.mlr.press/v28/arora13.pdf

Audickas, L., Hawkins, O., & Cracknell, R. (2017). UK Election Statistics: 1918-2017 (Briefing Paper No. CBP7529) (p. 89). London: House of Commons Library. Retrieved from http://researchbriefings.parliament.uk/ResearchBriefing/Summary/CBP-7529

Benoit, K. (2018). Quanteda: Quantitative Analysis of Textual Data. https://doi.org/10.5281/zenodo.1004683

Benoit, K., & Matsuo, A. (2018). Spacyr: Wrapper to the ’spaCy’ ’NLP’ Library. Retrieved from http://github.com/quanteda/spacyr

Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet Allocation. Journal of Machine Learning Research, 3(Jan), 993–1022.

Bligh, M., Merolla, J., Schroedel, J. R., & Gonzalez, R. (2010). Finding Her Voice: Hillary Clinton’s Rhetoric in the 2008 Presidential Campaign. Women’s Studies, 39(8), 823–850. https://doi.org/10.1080/00497878.2010.513316

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed). Hillsdale, N.J: L. Erlbaum Associates.

Fruchterman, T. M. J., & Reingold, E. M. (1991). Graph drawing by force-directed placement. Software: Practice and Experience, 21(11), 1129–1164. https://doi.org/10.1002/spe.4380211102

Gagolewski, M. (2018). R package stringi: Character string processing facilities. https://doi.org/10.5281/zenodo.1292492

Grimmer, J., & Stewart, B. M. (2013). Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts. Political Analysis, 21(03), 267–297. https://doi.org/10.1093/pan/mps028

Honnibal, M., & Montani, I. (2017). spaCy 2: Natural language understanding with Bloom embeddings, convolutional neural networks and incremental parsing. To Appear. Retrieved from https://spacy.io

Jones, J. J. (2016). Talk "Like a Man": The Linguistic Styles of Hillary Clinton, 1992-2013. Perspectives on Politics, 14(03), 625–642. https://doi.org/10.1017/S1537592716001092

Kelly, R., & White, I. (2016). All-women shortlists (Briefing Paper No. 5057) (p. 34). London: House of Commons Library. Retrieved from https://researchbriefings.parliament.uk/ResearchBriefing/Summary/SN05057

Kincaid, J. P., Fishburne, R. P., Rogers, R. L., & Chissom, B. S. (1975). Derivation of New Readability Formulas (Automated Readability Index, Fog Count and Flesch Reading Ease Formula) for Navy Enlisted Personnel: Fort Belvoir, VA: Defense Technical Information Center. https://doi.org/10.21236/ADA006655

Lee, M., & Mimno, D. (2014). Low-dimensional Embeddings for Interpretable Anchor-based Topic Inference. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 1319–1328). Doha, Qatar: Association for Computational Linguistics. https://doi.org/10.3115/v1/D14-1138

Mimno, D., Wallach, H. M., Talley, E., Leenders, M., & McCallum, A. (2011). Optimizing semantic coherence in topic models. In Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing (pp. 262–272). Edinburgh: Association for Computational Linguistics. Retrieved from https://dl.acm.org/citation.cfm?id=2145432.2145462

Newman, M. L., Groom, C. J., Handelman, L. D., & Pennebaker, J. W. (2008). Gender Differences in Language Use: An Analysis of 14,000 Text Samples. Discourse Processes, 45(3), 211–236. https://doi.org/10.1080/01638530802073712

Odell, E. (2018). Hansard Speeches and Sentiment V2.5.1 [dataset]. https://doi.org/10.5281/zenodo.1306964

Pennebaker, J. W., Boyd, R. L., Jordan, K., & Blackburn, K. (2015). The Development and Psychometric Properties of LIWC2015, 26. Retrieved from https://repositories.lib.utexas.edu/bitstream/handle/2152/31333/LIWC2015_LanguageManual.pdf

Pitkin, H. F. (1967). The concept of representation (1. paperback ed., [Nachdr.]). Berkeley, Calif.: Univ. of California Press.

Quinn, K. M., Monroe, B. L., Colaresi, M., Crespin, M. H., & Radev, D. R. (2010). How to Analyze Political Attention with Minimal Assumptions and Costs. American Journal of Political Science, 54(1), 209–228. https://doi.org/10.1111/j.1540-5907.2009.00427.x

Roberts, M. E., Stewart, B. M., & Airoldi, E. M. (2016). A Model of Text for Experimentation in the Social Sciences. Journal of the American Statistical Association, 111(515), 988–1003. https://doi.org/10.1080/01621459.2016.1141684

Roberts, M. E., Stewart, B. M., & Tingley, D. (2018). Stm: R Package for Structural Topic Models. Retrieved from http://www.structuraltopicmodel.com

Taddy, M. A. (2012). On Estimation and Selection for Topic Models. In Proceedings of the 15th International Conference on Artificial Intelligence and Statistics (Vol. 22, pp. 1184–1193). La Palma, Canary Island: JMLR W&CP.

Yu, B. (2014). Language and gender in Congressional speech. Literary and Linguistic Computing, 29(1), 118–132. https://doi.org/10.1093/llc/fqs073


  1. e.g. a reference to “the member for Bethnal Green and Bow” in keeping with Parliamentary convention of identifying MPs by their seat rather than their name would be followed by “(Rushnara Ali)”.

  2. Available online at: https://github.com/mikajoh/tidystm